dce/davideisinger.com
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

add links

Browse Source
This commit is contained in:
David Eisinger
2025-03-02 01:14:03 -05:00
parent 21c554c89b
commit 56c023648e
9 changed files with 3274 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

77
content/journal/dispatch-25-march-2025/index.md
View File

@@ -9,6 +9,38 @@ references:
url: https://harper.blog/2025/02/16/my-llm-codegen-workflow-atm/
date: 2025-03-02T05:57:54Z
file: harper-blog-l8lxlh.txt
- title: "A Taste of Vanlife"
url: https://prayash.io/journal/vanlife
date: 2025-03-02T06:12:05Z
file: prayash-io-vrlx4z.txt
- title: "I'm a feminist and I think it's harder to be a man than a woman."
url: https://celestemdavis.substack.com/p/hard-to-be-a-man
date: 2025-03-02T06:12:10Z
file: celestemdavis-substack-com-8yexkb.txt
- title: "Moving on from 18F. — Ethan Marcotte"
url: https://ethanmarcotte.com/wrote/leaving-18f/
date: 2025-03-02T06:12:11Z
file: ethanmarcotte-com-p78ctl.txt
- title: "The Tiny Book of Great Joys · Muffin Man"
url: https://muffinman.io/blog/the-tiny-book-of-great-joys/
date: 2025-03-02T06:12:13Z
file: muffinman-io-lxkzuw.txt
- title: "Ollama - NSHipster"
url: https://nshipster.com/ollama/
date: 2025-03-02T06:12:19Z
file: nshipster-com-b3vpys.txt
- title: "Is it okay?"
url: https://www.robinsloan.com/lab/is-it-okay/
date: 2025-03-02T06:12:23Z
file: www-robinsloan-com-dkgq9f.txt
- title: "Manton Reece"
url: https://www.manton.org/2025/02/09/i-missed-the-memo-for.html
date: 2025-03-02T06:12:30Z
file: www-manton-org-iaxj45.txt
- title: "You Can’t Post Your Way Out of Fascism"
url: https://www.404media.co/you-cant-post-your-way-out-of-fascism/
date: 2025-03-02T06:12:33Z
file: www-404media-co-3wvica.txt
---
Some thoughts here...
@@ -54,10 +86,43 @@ Some thoughts here...
### Links
* [Title][5]
* [Title][6]
* [Title][7]
* [A Taste of Vanlife][5]
[5]: https://example.com/
[6]: https://example.com/
[7]: https://example.com/
> Camping has always been one of my favorite hobbies, but I've always wondered what life might be like on four wheels. It's difficult to think about living on the road indefinitely—and understandably so. I'm not really into the idea of getting rid of all of my posessions and living out of a van, constantly on the move. But surely, there's a way to try out that lifestyle without making such a drastic commitment?
* [I'm a feminist and I think it's harder to be a man than a woman.][6]
> Very earnestly I believe that despite greater access to power and resources, the box labeled “socially acceptable ways to be a man” is much smaller than the box labeled “socially acceptable ways to be a woman.”
* [Moving on from 18F. — Ethan Marcotte][7]
> During that time, a friend suggested that while things were calm at work, I should write down some lines I wouldn’t want to cross: things I’d want to watch out for that, if they materialized, might be a reason to leave. This was wonderful advice, and I’m grateful to them for it. Equipped with a plan, even a small one, I started thinking through what my lines would be.
* [The Tiny Book of Great Joys · Muffin Man][8]
> If you are interested in how I over-engineered the process of making a tiny book for my wife, using AI, a pen plotter, a 3D printer, and a lot of time, you are in the right place. The book is titled The Tiny Book of Great Joys, and here is how it turned out.
* [Ollama - NSHipster][9]
> If you wait for Apple to deliver on its promises, you’re going to miss out on the most important technological shift in a generation. The future is here today. You don’t have to wait. With Ollama, you can start building the next generation of AI-powered apps right now.
* [Is it okay?][10]
> How do you make a language model? Goes like this: erect a trellis of code, then allow the real program to grow, its development guided by a grueling training process, fueled by reams of text, mostly scraped from the internet. Now. I want to take a moment to think together about a question with no remaining practical importance, but persistent moral urgency: Is that okay?
* [I missed the memo for the outside space at this coffee shop… – Manton Reece][11]
> The bittersweet irony with parenting is not knowing until years later what you had.
* [You Can’t Post Your Way Out of Fascism][12]
> We don’t need any more irony-poisoned hot takes or cathartic, irreverent snark. We need to collectively decide what kind of world we actually do want, and what we’re willing to do to achieve it.
[5]: https://prayash.io/journal/vanlife
[6]: https://celestemdavis.substack.com/p/hard-to-be-a-man
[7]: https://ethanmarcotte.com/wrote/leaving-18f/
[8]: https://muffinman.io/blog/the-tiny-book-of-great-joys/
[9]: https://nshipster.com/ollama/
[10]: https://www.robinsloan.com/lab/is-it-okay/
[11]: https://www.manton.org/2025/02/09/i-missed-the-memo-for.html
[12]: https://www.404media.co/you-cant-post-your-way-out-of-fascism/

504
static/archive/celestemdavis-substack-com-8yexkb.txt Normal file
View File

@@ -0,0 +1,504 @@
[1]
Matriarchal Blessing
[2]Matriarchal Blessing
SubscribeSign in
Share this post
[8]
[https]
Matriarchal Blessing
Matriarchal Blessing
I'm a feminist and I think it's harder to be a man than a woman.
Copy link
Facebook
Email
Notes
More
I'm a feminist and I think it's harder to be a man than a woman.
Systemically women have it worse, but behaviorally, men are more limited.
[9]
[htt]
[10]Celeste Davis
Sep 15, 2024
881
Share this post
[12]
[https]
Matriarchal Blessing
Matriarchal Blessing
I'm a feminist and I think it's harder to be a man than a woman.
Copy link
Facebook
Email
Notes
More
[13]
155
201
[14]
Share
Article voiceover
1×
0:00
-9:44
Audio playback is not supported on your browser. Please upgrade.
When men say it is hard to be a man, I fully believe them.
I think it is very, very hard to be a man.
Perhaps since I write an essay about patriarchy each week, you may think I’m
saying that in a sarcastic voice “Oh it’s SOOoooOO hard to have society built
just for you[15]1 and have all the power in government, business and religion
for *checks watch* all of recorded history.[16]2 Boo hoo. Sad violins for you.”
But I’m not being sarcastic.
Very earnestly I believe that despite greater access to power and resources,
the box labeled “socially acceptable ways to be a man” is much smaller than the
box labeled “socially acceptable ways to be a woman.”
[17]
[https]
The masculine is at the top of the power hierarchy, but their behavior is more
limited.
Men’s behavior is limited to the masculine, but women have access to both
masculine and feminine behavior.
[18]
[https]
[19]
[https]
Women are not allowed to LOOK masculine, but they are often rewarded for ACTING
masculine- being strong and assertive, being one of the guys, favoring music,
books and media by men, ridiculing girly things, etc.
I’m going to paint with broad strokes here with obvious exceptions, but in
generalized terms of social acceptability:
• Women can be strong, but men can’t be weak.
• Women can wear blue, but men can’t wear pink.
• Girls can have boy names, but boys can’t have girl names.
• Women can be executives, but men can’t be homemakers.
• Girls can play basketball, but boys can’t do ballet.
• Women read books by men, but men don’t read books by women.
• Little girls can play with trucks and dinosaurs, but little boys can’t play
with dolls.
• Women can lift weights, but men don’t do pilates.
• Women can like action movies, but men can’t like chick flicks.
• Women can act tough and assertive, but men can’t act soft and submissive.
[20]
Strong Girl GIFs | Tenor
Women have access to the whole range of human behavior; men only have access to
the masculine.[21]3
(And yes, [22]blurring the gender binary makes everything better for everyone.)
Why is this?
Do you have a guess? I’ll give you a clue. Starts with a p. Rhymes with
schmatriarchy.
It is because with live in a system that values and rewards all things man.
Men have more power, but are more limited in how they can act.
Speaking on this predicament of [24]patriarchy harming men,
[25]Lane Anderson
recently quoted researcher Michael Kimmel who said,
“Men are in power as a group but do not feel power as individuals.. Men
were raised to believe themselves entitled to feel that power, but do not
feel it. No wonder men are frustrated and angry...
Failure to embody the rules of manhood is a source of men's confusion and
pain... it is unrealizable for any man. But we keep trying, vainly, to
measure up. American masculinity is a relentless test."
Thomas Page McBee speaks to this juxtaposition of men having more power but
less connection in his book [26]Amateur.
Thomas is a trans man. Since transitioning he has noticed that when he speaks,
everyone is quiet. No one interrupts him anymore. He says, “It was wonderful
and weird. Until I was a man, I had no idea how good men have it at work.”
[27]
amateur.jpg
But there are downsides. A year after his transition Thomas lost his mother. He
didn’t realize how much he needed human touch until it disappeared from his
life. No one hugged him anymore. He could go months without anyone touching
him. This was a foreign experience to Thomas.
He says he has become hyper aware of both how acceptable his anger is, and how
unacceptable his sadness is. The people in his life- men and women alike- are
visibly uncomfortable when he cries.
He misses the deep friendships that came easily as a woman.
It's hard to make friends when all the ingredients necessary to
friendship--vulnerability, compassion and thoughtfulness-- are seen as a threat
to masculinity.
[28]
Thomas Page McBee by Julie Greicius
Thomas Page Mcbee. [29]source
 Recently
[30]15thCenturyFeminist
interviewed
[31]Jeremy Mohler
about his experience of performing masculinity. Of his teenage years he said,
“I couldn't admit that I wanted to be closer with my friends. That would
mean I was 'soft,' 'girly,' or 'gay.' I had to act stoic, like everything
was fine, even if I was starving for love and connection inside."
Jeremy and Thomas are far from the only men having trouble with connection.
Male loneliness is being called an epidemic. [32]Psychologist Nick Norman
describes the problem in this way:
"Men have often reported having fewer friends and social connections to
rely on, [33]with 15 percent saying they have no close friends at all. Yet,
when surveyed, men often report wanting more fulfilling relationships. What
is keeping men from these connections when it’s such a fundamental need?...
the issue lies in the unspoken rules men are handed in boyhood."
The unspoken rule #1? You are not allowed to act like a girl.
Unfortunately friendships, relationships and human connection require all sorts
of girly things like empathy, showing emotion, being vulnerable and active
listening.
Men report wanting more fulfilling relationships, but they have been barred
from the behaviors that fulfilling relationships require.
“Learning to wear a mask is the first lesson in patriarchal masculinity a boy
learns. He learns that his core feelings cannot be expressed if they do not
conform to the acceptable behaviors sexism defines as male. Asked to give up
the true self in order to realize the patriarchal ideal, boys learn
self-betrayal early and are rewarded for these acts of soul murder.”  - bell
hooks [34]the will to change
If men are in charge like patriarchy says, why are men suffering so much?
Many take the fact that it is difficult to be a man to mean that patriarchy is
not real—
Men are the victims of society, so how could they be the instigators of their
own victimhood? It makes no sense. If men set this system up to favor men,
wouldn’t men be thriving? If patriarchy is real, why are men suffering?
Great question. I’m going to let the brilliant creator and thinker [36]Cyzor
[37]4 cover this one.
He received a message complete with two pages of documentation showing that men
are the real victims (more suicide, more likely to be victims of crime, more
likely to be charged “guilty” in court, etc) so all this whining about
patriarchy is bullshit.
Here is Cyzor’s response:
[38][https]
[39]@cyzorgg[40]Day 12 and the final day of reposting my top. videos of the
year. This is by far both my favorite video and one of the best performing that
I made this year and Im so glad everyone liked it and felt it was helpful and
useful for them in their own lives. In the next year I plan to expand and grow
as a content creator and try tons of new avenues, work with brands, and see
what works for me. but whatever I do, I'll continue to share my thoughts in
hopes they help others! I appreciate you all for the love and support this
year!!
[alert-circ]Tiktok failed to load.
Enable 3rd party cookies or use another browser
“Men are more likely to die by homicide. Men are more likely to be victims
of crime. Men are also more likely to COMMIT all of those crimes. 72% of
crimes are done by men. Men are the victims of other men.
Men are discriminated in universities, that are run by men. Men are going
to jail longer in a justice system, that is run by men. Bro. You are mad at
the same structure that women are trying to get rid of.
If you are complaining about court policies, social expectations, crime-
bro - you are complaining about the patriarchy.” - Cyzor
Men suffer under patriarchy because patriarchy is a system set up by a few men
over everyone else- including most men. It is a system that allows for the most
greedy, aggressive and selfish to rise to the top and dominate over everyone
else.
DNA scientists have discovered that 8000 years ago on average [41]only one man
passed on his DNA for every 17 women. Not just in one place, but as a global
average. The majority of men were not procreating. Not because there was a mass
death of males, but because only a few men accumulated wealth, power and women
and withheld resources from the rest.
A small minority of men were allowed to rule and make life worse for everyone
else- men and women alike.
[42]And as we discussed last week with lawns, just because something 1. begins
for no reason and 2. harms everyone does not mean we won’t religiously devote
ourselves to that thing.
[43]
Ancient Egyptian Battles & Wars - Egypt Tours Portal (US)
So manly.
We often hear that patriarchy hurts men too, but how many men actually believe
this?
All too often, once the p word enters the chat, defenses are raised,
conversations are ended.
But Cyzor, no stranger to patriarchy pushback, also helped me to understand why
it is that [44]so many men shut down and become defensive when patriarchy is
brought up when he said,
“I criticize those aspects of masculinity that I don’t think are healthy
for men and if you’ve internalized those aspects of masculinity as being
part of you, it’s going to feel like I am criticizing you.”
Society trains men to put manhood at the very center of their identity. When we
talk about patriarchy, we talk about those things which men have built their
identities and lives around. Of course they are defensive. It feels like an
attack.
But we have to stop conflating patriarchy and men.
Our ability to have really important conversations relies on our ability to
make this distinction.
So often those of us who bring up patriarchy and those who shut it down are on
the same side. We want the same things. We want life to be better for men.
We must make room for the reality that we talk about patriarchy not because we
hate men. We talk about patriarchy because we love men.
We have/are sons, husbands, fathers, brothers, friends. And we see the small
box patriarchy has handed to men.
To #smashthepatriarchy is not to smash men- it is to smash this system that
makes men miserable. Smash the too-small box that is making men lonely and sick
and estranged from the full range of their humanity.
I’ll echo the [45]words of Cyzor to close us out today:
"I want men to succeed. I envision a world where men are allowed to
experience the full range of human emotion, where they are not shamed for
crying or being vulnerable, where they are not shamed if they are not
physically tall or strong or rich… where men can have real emotional
connection to one another, have a real support system with people who lift
them up and help them when they are down, where men aren't under the
impression that masculinity means experiencing the trials of life by
yourself."
Amen.
It IS hard to be a man. It is very hard.
It would be a whole lot easier without patriarchy.
[46]Leave a comment
[47]1
If you’re white
[48]2
If you’re white
[49]3
Yes there is some social stigma and repercussions for women acting masculine
(in the dating pool, at work), but not to the extent of the social punishment
for men acting feminine.
[50]4
[51]Follow him immediately. His videos are so healing.
Matriarchal Blessing is a reader-supported publication. To receive new posts
and support my work, consider becoming a free or paid subscriber.
[62][ ]
Subscribe
Thanks for reading Matriarchal Blessing! This post is public so feel free to
share it.
[64]Share
881
Share this post
[66]
[https]
Matriarchal Blessing
Matriarchal Blessing
I'm a feminist and I think it's harder to be a man than a woman.
Copy link
Facebook
Email
Notes
More
[67]
155
201
[68]
Share
Discussion about this post
CommentsRestacks
[ht]
[ ]
[ ]
[ ]
[ ]
[72]
[ht]
[73]Arturo Mijangos
[74]Sep 15
Liked by Celeste Davis
This is why I continue to champion the dismantling of patriarchy. I will
showcase what accessing feminine traits looks like. I will continue to teach
how patriarchy impacts and limits men. I will continue to support lgbtq+ rights
who exemplify broader expressions of the gender spectrum.
Expand full comment
Reply
Share
[77]12 replies by Celeste Davis and others
[78]
[ht]
[79]Kimberley Healey
[80]Sep 15
Liked by Celeste Davis
Absolutely! Thanks for the tiktok repost - he nails it. When I taught Women's
Lit I brought in a bunch of my male colleagues to explain their experience with
masculinity. Football coach, older teachers, younger ones - we all changed that
day when they started sharing things they had NEVER told anyone. The
loneliness, lack of touch and constant pressure to PROVE they were actually men
were heartbreaking. It does suck to be a woman but I get to just be one, I
don't have to step up and BE a man in the same way that men have to do in the
US. One book that really helped me teach my course was For the Love of Men by
Liz Plank. Very curious to see the comments on this great essay!
Expand full comment
Reply
Share
[83]6 replies by Celeste Davis and others
[84]153 more comments...
TopLatestDiscussions
No posts
Ready for more?
[99][ ]
Subscribe
© 2025 Celeste Davis
[101]Privacy ∙ [102]Terms ∙ [103]Collection notice
[104] Start Writing[105]Get the app
[106]Substack is the home for great culture
Share
Copy link
Facebook
Email
Notes
More
This site requires JavaScript to run correctly. Please [108]turn on JavaScript
or unblock scripts
References:
[1] https://celestemdavis.substack.com/
[2] https://celestemdavis.substack.com/
[8] https://substack.com/home/post/p-148892931?utm_campaign=post&utm_medium=web
[9] https://substack.com/@celestemdavis
[10] https://substack.com/@celestemdavis
[12] https://substack.com/home/post/p-148892931?utm_campaign=post&utm_medium=web
[13] https://celestemdavis.substack.com/p/hard-to-be-a-man/comments
[14] javascript:void(0)
[15] https://celestemdavis.substack.com/p/hard-to-be-a-man#footnote-1-148892931
[16] https://celestemdavis.substack.com/p/hard-to-be-a-man#footnote-2-148892931
[17] https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F94798dee-78fc-43c9-967d-664b966d85d7_1456x1048.png
[18] https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F41c6b879-2f48-4de6-b3e4-500e9915e9d3_1456x1048.png
[19] https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F126de46d-8e49-4ca8-85c1-7ae5edbebe01_1456x1048.png
[20] https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd9e78a3d-537f-47b1-b9cb-60a6d548367d_220x143.gif
[21] https://celestemdavis.substack.com/p/hard-to-be-a-man#footnote-3-148892931
[22] https://substack.com/@celestemdavis/p-146095286
[24] https://matriarchyreport.substack.com/p/patriarchy-is-bad-for-men-and-boys
[25] https://open.substack.com/users/1628200-lane-anderson?utm_source=mentions
[26] https://www.thomaspagemcbee.com/amateur
[27] https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc7ef7faf-a030-4e37-a271-03b1cd2518d3_300x478.jpeg
[28] https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F29bf2530-4eff-49c7-8c55-6d3dc2151c67_3008x2000.jpeg
[29] https://midnightbreakfast.com/thomas-page-mcbee
[30] https://open.substack.com/users/129144801-15thcenturyfeminist?utm_source=mentions
[31] https://open.substack.com/users/3962129-jeremy-mohler?utm_source=mentions
[32] https://www.psychologytoday.com/us/blog/mens-mental-health-matters/202301/why-men-are-lonelier-than-ever
[33] https://www.americansurveycenter.org/why-mens-social-circles-are-shrinking/
[34] https://www.secondsale.com/p/the-will-to-change-men-masculinity-and-love/622949?ean13=9780743456081&id_product_attribute=58734820&campaignid=21540511568&adgroupid=&keyword=&device=c&gad_source=1&gclid=CjwKCAjw6JS3BhBAEiwAO9waF9yLlP1Epl1Fd8T6ro4IQPtiay3SjgeYx_BK8R1ydqPJ2klvbg5dChoCNYAQAvD_BwE
[36] https://www.tiktok.com/@cyzorgg
[37] https://celestemdavis.substack.com/p/hard-to-be-a-man#footnote-4-148892931
[38] https://www.tiktok.com/@cyzorgg/video/7318190925937675562
[39] https://www.tiktok.com/@cyzorgg
[40] https://www.tiktok.com/@cyzorgg/video/7318190925937675562
[41] http://www.psmag.com/nature-and-technology/17-to-1-reproductive-success
[42] https://substack.com/@celestemdavis/p-148630300
[43] https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F3e3de959-b86a-4c5a-9260-1554fd1b3fee_2048x1400.jpeg
[44] https://www.tiktok.com/@cyzorgg/video/7317746877271706922
[45] https://www.tiktok.com/@cyzorgg/video/7317746877271706922
[46] https://celestemdavis.substack.com/p/hard-to-be-a-man/comments
[47] https://celestemdavis.substack.com/p/hard-to-be-a-man#footnote-anchor-1-148892931
[48] https://celestemdavis.substack.com/p/hard-to-be-a-man#footnote-anchor-2-148892931
[49] https://celestemdavis.substack.com/p/hard-to-be-a-man#footnote-anchor-3-148892931
[50] https://celestemdavis.substack.com/p/hard-to-be-a-man#footnote-anchor-4-148892931
[51] https://www.tiktok.com/@cyzorgg
[64] https://celestemdavis.substack.com/p/hard-to-be-a-man?utm_source=substack&utm_medium=email&utm_content=share&action=share
[66] https://substack.com/home/post/p-148892931?utm_campaign=post&utm_medium=web
[67] https://celestemdavis.substack.com/p/hard-to-be-a-man/comments
[68] javascript:void(0)
[72] https://substack.com/profile/165554130-arturo-mijangos?utm_source=comment
[73] https://substack.com/profile/165554130-arturo-mijangos?utm_source=substack-feed-item
[74] https://celestemdavis.substack.com/p/hard-to-be-a-man/comment/69084607
[77] https://celestemdavis.substack.com/p/hard-to-be-a-man/comment/69084607
[78] https://substack.com/profile/6902565-kimberley-healey?utm_source=comment
[79] https://substack.com/profile/6902565-kimberley-healey?utm_source=substack-feed-item
[80] https://celestemdavis.substack.com/p/hard-to-be-a-man/comment/69082532
[83] https://celestemdavis.substack.com/p/hard-to-be-a-man/comment/69082532
[84] https://celestemdavis.substack.com/p/hard-to-be-a-man/comments
[101] https://substack.com/privacy
[102] https://substack.com/tos
[103] https://substack.com/ccpa#personal-data-collected
[104] https://substack.com/signup?utm_source=substack&utm_medium=web&utm_content=footer
[105] https://substack.com/app/app-store-redirect?utm_campaign=app-marketing&utm_content=web-footer-button
[106] https://substack.com/
[108] https://enable-javascript.com/

466
static/archive/ethanmarcotte-com-p78ctl.txt Normal file
View File

@@ -0,0 +1,466 @@
[1]Skip to content
Site navigation
• [2]Home
• [3]Journal
• [4]Books
• [5]Work
• [6]Contact
[7] Search
[8] Ethan Marcotte’s homepage Posted on 18 February 2025
Moving on from 18F.
Note: This post gets into the last few weeks of American politics. If that’s
not your cup of tea, or if that’s a stressful topic for you, please feel free
to skip this one. (Also, it’s a bit long. Sorry about that.)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Last week, I finished my tenure as [9]a designer at 18F.
I want to state up front: I’m not leaving under a “[10]deferred resignation.” I
also wasn’t laid off. (Though it’s possible I almost was; more on that later.)
Instead, I resigned from my position as a product designer, submitting two
weeks’ notice…well, two weeks ago.
Before I get into any of that, I’d like to write a bit about 18F, and why it
was so hard to leave.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
While I was writing this post, I thought I’d revisit [11]what I wrote when I
joined 18F last May:
1. Every single person I’ve met this week — and I’ve met quite a few!
 — has been smart, kind, and really happy to be working where they do.
As someone new to the organization, that’s so encouraging to see.
2. It’s, like, remarkably energizing to be around people who are really
(really, really) passionate about making digital services work better
for people.
Honestly, that holds up. Because really, the thread here is the people working
at [12]18F, and the culture they’d built: I really, really liked showing up for
work each morning. Everyone I met at 18F was inviting and kind, and excited to
talk about what they were working on. (And just as crucially, what they did
outside work.)
And my goodness, they were helpful — which, as a new kid joining the team, I’m
always going to remember. Here’s one example: during my first month, I was
grousing about some weird little computer issue, and a random coworker just
offered to hop on a call to look at it with me. They hadn’t dealt with the
issue before, and they definitely hadn’t dealt with me before, but they thought
they might help a coworker out. And that impulse — maybe I can help someone out
 — sums up so many of my interactions with everyone at 18F. They were, and are,
a remarkable group of people.
At the same time, I was proud of the work I was doing. Alongside my coworkers
at 18F, I worked with client teams to help them define requirements, refine
their designs, and build better products. I even got asked to pitch in on a
small branding project, and I’d be the last person to call myself a brand
designer. But I mention that because I was often asked to stretch myself, and
every single time I felt safe trying something new — safe, and supported by my
team. I can count on one hand the number of times over my career that I’ve felt
that kind of safety at work. I doubt that’s true of every job in government,
but I know it was true for me at 18F.
I know it sounds pat, but 18F was one of the best places I’ve ever worked.
Until it wasn’t, and I felt I had to leave.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Before I dive in, here are a couple points that’ll become relevant:
• I was considered a probationary employee because I’d been employed by the
government for less than a year. [13]Probationary employees don’t have most
of [14]the protections afforded to “full” employees, and can be dismissed
more easily.
• Due to some idiosyncrasies of how our roles were defined, many (most?)
people in my organization were legally not eligible to join a union.
So. After last year’s election, I was trying to decide whether or not I could
stay at the job. A far-right candidate had won the election^[15]1, and was
threatening to [16]reshape the government into something more partisan, more
regressive, and more autocratic. My job involved putting rectangles on screens,
and couldn’t have been further from any kind of political influence or impact.
But despite that, I didn’t know if I could let myself be part of that
government, even in a small way. (Also, [17]as you might have guessed: I was
panicking.)
During that time, a friend suggested that while things were calm at work, I
should write down some lines I wouldn’t want to cross: things I’d want to watch
out for that, if they materialized, might be a reason to leave. This was
wonderful advice, and I’m grateful to them for it. Equipped with a plan, even a
small one, I started thinking through what my lines would be.
I’ll spare you the whole list, but I’ll share three of the entries.
1. First, I need to work remotely. If the incoming administration made good on
its promise to end teleworking for federal workers, I’d likely have to find
another job. (This is, of course, [18]why “return to office” policies
happen.)
2. The second line was whether I’d be asked to work on a project that could
kill or surveil people. I know precisely what governments are capable
of — for good and for ill. But one of the things that drew me to the work
at 18F was that I understood they tried to weigh individual workers’
preferences when projects were staffed. I figured if that ever changed, and
I was asked to work on something I was morally opposed to, it’d be time to
leave.
3. The third was being asked to meet with someone who didn’t work for the
government, and being asked to discuss what I did for work.
The first two were things I looked into when I was first interviewing at 18F:
some of the basic criteria I was screening potential employers for. The third
was driven at least in part by the election, and by the billionaire they were
putting in charge of “government tech modernization.” I expected that if things
went south, he’d just try to run the same horrible [19]Twitter layoffs handbook
, and bring in employees from his other companies to rank — and cull — workers.
But it wasn’t just about that. Many things started happening to the federal
government after the inauguration, none of them good. While the administration
was conducting its vicious rollback of civil liberties and publicly funded
research, [20]this billionaire’s so-called “department” was sweeping through
[21]various federal agencies, pushing aside career civil servants and the law
to [22]hoover up [23]radioactively [24]sensitive data — our data, bought and
paid for with our tax dollars, I should add.^[25]2 And from what I’d read the
group was acting on [26]dubious legal authority, and with even less [27]
oversight or [28]transparency. I didn’t want to sit down with anyone involved
in that, and pretend like any part of their work was lawful, legitimate, or
moral.
Anyway. The list was a tremendous help; I’ll always be grateful to the friend
who suggested it. But given the speed at which government typically moves, I
assumed I’d have several months before I’d have to wrestle with any of these
questions. If not longer.
(I know, I know. I’m in the future, too.)
A few weeks ago, a member of [29]the new leadership announced they’d be
reaching out to workers to discuss their recent “technical wins”, in order to
better understand how the organization worked. The stress on “technical wins”
to a [30]cross-functional organization felt significant to me; it also felt
significant that most of the sessions seemed to be getting scheduled with folks
who’d only recently joined government — probationary employees.
Just to state the obvious, this isn’t what you do when you want to understand
how your organization works; it is what you do when you’re preparing to slash
the size of your workforce. As you might imagine, this caused no small amount
of panic across the agency, including within 18F. The new leadership hadn’t
communicated these plans to anyone before making their announcement, which left
18F’s own leaders and supervisors frantically working to fill in the
information void.
Shortly after the announcement, I started hearing about folks who’d had their
meetings, but that they didn’t meet with the director who said they’d be
conducting the interviews. Instead, they found themselves on a call with people
who wouldn’t say where they worked in government; in a few cases, some people
wouldn’t disclose their last names, or any part of their names.
And while I was watching these reports trickle in, I got a calendar invitation
for my own interview. From the first email announcing the meetings, I figured
one of my lines was in danger of being crossed; I just figured I’d have more
time.
With only a few hours before my interview, I did a quick overview of my
options. It looked like this:
1. I could do the interview.
2. I could refuse to do the interview.
3. I could delay: call out sick, take a personal day, whatever.
4. I could resign.
The first item wasn’t really an option, as sitting down with this “department”
wasn’t something I could let myself do. Refusing to participate would’ve likely
been seen as insubordination by a probationary hire; delaying would’ve just
been, well, delaying the inevitable. (It also could have been seen as
insubordination.) My math would’ve been different if I wasn’t probationary or,
even better, if I’d been allowed to join a union. But given my lack of labor
protections, and the options available before me, leaving 18F — withholding my
labor — felt like my best and only option. I called a meeting with my
supervisor, and gave two weeks’ notice.
In a terrible coda, a large number of [31]probationary employees were summarily
let go at [32]my agency just before my last day.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Leaving was the right call for me, but I’ll never feel good about the decision.
I mean, there’s the grief angle: up until about a month ago, I was working on
projects that felt like they mattered, and working alongside people who cared
about helping government services work better for the public. A few months ago,
I would’ve told you I’d like to stay there for years, which is not something
I’ve said about any other place I’ve ever worked. I am incredibly sad to leave
this job.
And look, being able to leave is, flatly, a privileged option: I can’t not work
forever, but I can not work for a little bit. Most of my coworkers didn’t have
that option. Some had just bought a house; some returned from parental leave,
only to learn they might be losing the jobs they’d counted on to support their
families.
I’m also angry at what was taken from me. At what’s being taken from all of us.
I’ve watched a wonderful job, a wonderful place to work, a wonderful team get
pulled apart by rich men in ill-fitting suits, each of them parroting the same
talking points around “realignment” and “right-sizing”.^[33]3
But what’s happening right now is not about “government efficiency,” nor is it
about “cost-cutting.” I would gently urge you to look at the net worth of the
people who are telling you otherwise. After all, there is no financial
analysis; no review of possible downsides, no weighing of potential negative
impacts. There is no discussion of what could happen if our math is wrong? Or
even more importantly, no consideration of who might be harmed?
Instead, as [34]Anil Dash predicted, the billionaire’s so-called “efficiency”
“department” is best understood as a sprawling form of [35]procurement capture,
in which a group of impossibly rich individuals are trampling over the
regulations — and the federal workers — that stand between them and a deep,
deep [36]revenue [37]stream: [38]your tax dollars. And as they do, they’re
making an explicitly fascist move to roll back rights for every marginalized
community in the country — for anyone who doesn’t look like them, or who stands
in their way.
So, yes. This is a wholesale attack on the American safety net, led by
billionaires and far-right politicians who are frighteningly comfortable with
fascism and autocracy. The last month has been called a coup by [39]politicians
, [40]researchers, and [41]watchdogs alike. I don’t want to diminish the harm
these people will do — the harm they are doing. I also don’t want to downplay
the terror of this moment, because lord knows I fucking feel it.
At the same time: what’s happening right now is also a labor story.
If the American government is slow-moving, it’s because rapid change is deadly
when you’re talking about healthcare, social security checks, market
regulations, food safety, or any of the other countless critical functions it
performs. Those federal agencies are, quite simply, infrastructure. And as [42]
Deb Chachra showed in [43]her excellent book, infrastructure is how a society
invests in its future: in its ongoing economic, societal, and political
stability.
In government, that infrastructure is built by laws, policies, and regulations.
But regulations alone do not infrastructure make. Regulations require workers
to become infrastructure: those workers who labor to understand new policies,
how best to enact them, and then work to make them legible and understandable
to the American public — and, yes, to enforce them. Without those federal
workers, and their labor, these systems fall apart. And the architects of this
assault on the federal workforce are keenly aware of that fact.
The last month has, flatly, been hell. But even so, I wouldn’t trade away my
time at 18F for anything. It was a fantastic place to work, filled with
genuine, hard-working people who cared for that work and for each other. Even
when things got rough, I saw the leaders of 18F scramble to answer their team’s
questions; I saw coworkers reaching out to support each other in countless
little ways. All while ensuring they got their project work in on time. I saw
something wonderful at work, in my work. I’m always going to be grateful for
that, and to my coworkers.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Resources
If this story’s moved you, I hope it moves you to action. Because the workers I
mention above quite literally need your support.
A few resources, if you’re interested:
• Wired has some [44]good coverage on the layoffs I described above, and [45]
on the billionaire coup more generally.
• [46]Labor Notes also has some indispensable coverage around [47]this
administration’s attacks on the federal workforce, and how organized labor
is fighting back.
• The [48]Working Families Party and [49]Emily Amick both had some great
primers on what it means to call your members of Congress, if that’s a
thing you’re able to do.
• If you’re looking for other ways to get engaged, [50]Mariame Kaba has
pulled together a massive list of [51]actions that are not protesting or
voting.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Footnotes
1. A victory by the slimmest of margins, mind you. But still a victory. [52]↩︎
2. And, seemingly coincidentally, thereby ending [53]various investigations
against the head of said “department”, and occasionally [54]lining his
pockets. [55]↩︎
3. And perhaps just as excruciating for me: “datalake”. [56]↩︎
Tagged with
• [57]work
• [58]jobs
• [59]politics
• [60]us politics
• [61]employment
• [62]government
Related posts
• [63]On context. I read these two essays some time ago, and I keep returning
to them. I bet you’ll like them too.
• [64]The bricks we lay. Design is not neutral.
• [65]Free, faster. Many of the free web themes I’ve seen recently are…slow.
How can we fix that?
• [66]Hello, Editorially. I’ve cofounded a startup with some dear friends.
It’s called Editorially. I’d like to tell you a little about it.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
You can find more entries in [67]my journal.
Read another post
[68] Previously: A challenge of blog questions.
What did I just read?
Photo of Ethan standing in front of a leafy green hedge.
Hi! I’m Ethan Marcotte, an independent web designer and writer. Some time ago,
I coined the term “responsive web design.” (You can [69]read more about me or
[70]my work, if you like.)
My latest book
[book-ydatu]
[71]You Deserve a Tech Union is a book about the tech industry’s resurgent
labor movement, and how you can—and should—be part of it. [72]Learn more.
Subscribe for updates!
If you enjoyed this post, sign up to get new journal entries emailed to you:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Your email address:
[73][ ]
[74][ ]
[75][Subscribe]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Site footer
[76] Ethan Marcotte’s homepage
Here
• [77]Home
• [78]Journal
• [79]Books
• [80]Work
• [81]Contact
Elsewhere
• [82]Mastodon
• [83]Bluesky (sorta)
• [84]LinkedIn (reluctantly, semi-ironically)
• [DEL:Instagram (occasionally):DEL]
• [DEL:Twitter:DEL]
Copyright © 1999–2025 Ethan Marcotte. All rights reserved.
No part of this website may be used or reproduced in any manner for the purpose
of training artificial intelligence technologies or systems. The author
expressly reserves this work from the text and data mining exception.
(That means get lost, “AI” scrapers.)
[85]Accessibility statement. [86]RSS feed. [87]Back to top.
[88]Skip to content
Current page
Search:
[93][ ]
References:
[1] https://ethanmarcotte.com/wrote/leaving-18f/#content
[2] https://ethanmarcotte.com/
[3] https://ethanmarcotte.com/wrote/
[4] https://ethanmarcotte.com/books/
[5] https://ethanmarcotte.com/work/
[6] https://ethanmarcotte.com/contact/
[7] https://ethanmarcotte.com/search/
[8] https://ethanmarcotte.com/
[9] https://ethanmarcotte.com/wrote/18f/
[10] https://www.afge.org/article/afge-cautions-feds-not-to-be-tricked-into-resigning-you-might-not-get-paid/
[11] https://ethanmarcotte.com/wrote/18f/
[12] https://en.wikipedia.org/wiki/18F
[13] https://federalnewsnetwork.com/workforce-rightsgovernance/2025/02/what-are-the-rules-for-probationary-periods-and-federal-employees/
[14] https://www.mspb.gov/
[15] https://ethanmarcotte.com/wrote/leaving-18f/#fn-margins
[16] https://en.wikipedia.org/wiki/Project_2025
[17] https://ethanmarcotte.com/wrote/catalog/
[18] https://finance.yahoo.com/news/quarter-bosses-admit-return-office-104103939.html
[19] https://web.archive.org/web/20221102222024/https://www.washingtonpost.com/technology/2022/10/29/elon-musk-twitter-takeover/#:~:text=The%20note%20continued%3A%20%E2%80%9CPlease%20come%20prepared%20with%20code%20as%20a%20backup%20to%20review%20on%20your%20own%20machines%20with%20Elon.%E2%80%9D%20Later%2C%20people%20inside%20the%20company%20reported%20that%20Tesla%20engineers%20were%20in%20fact%20reviewing%20the%20code.
[20] https://www.bbc.com/news/articles/c23vkd57471o#:~:text=Despite%20its%20full%20name%2C%20Doge%20is%20not%20an%20official%20government%20department%2C%20which%20would%20have%20had%20to%20be%20established%20by%20an%20act%20of%20Congress.
[21] https://www.wired.com/story/elon-musk-lackeys-office-personnel-management-opm-neuralink-x-boring-stalin/
[22] https://abcnews.go.com/US/judge-decide-block-doge-accessing-sensitive-labor-department/story?id=118575362
[23] https://www.nbcnews.com/politics/doge/doge-affiliated-employee-accessed-irs-system-sensitive-taxpayer-inform-rcna192423
[24] https://www.cnn.com/2025/02/17/politics/doge-irs-taxpayer-data/index.html
[25] https://ethanmarcotte.com/wrote/leaving-18f/#fn-conflicts
[26] https://www.washingtonpost.com/business/2025/02/04/elon-musk-government-legal-doge/
[27] https://oversightdemocrats.house.gov/sites/evo-subsites/democrats-oversight.house.gov/files/evo-media-document/2025-02-06.Dem%20Members%20to%20IGs%20re%20Musk.pdf
[28] https://www.404media.co/doge-employees-ordered-to-stop-using-slack-while-agency-transitions-to-a-records-system-not-subject-to-foia/
[29] https://www.gsa.gov/about-us/newsroom/news-releases/gsa-announces-new-commissioners-tts-director-and-general-counsel-01242025
[30] https://experience.dropbox.com/resources/cross-functional-teams
[31] https://www.npr.org/2025/02/13/nx-s1-5296928/layoffs-trump-doge-education-energy
[32] https://fedscoop.com/gsa-looks-to-terminate-probationary-employees/
[33] https://ethanmarcotte.com/wrote/leaving-18f/#fn-datalake
[34] https://www.anildash.com/
[35] https://www.anildash.com/2025/01/04/DOGE-procurement-capture/
[36] https://newrepublic.com/article/191506/musk-bezos-pichai-zuckerberg-microsoft-trump-climate
[37] https://www.technologyreview.com/2024/12/04/1107897/openais-new-defense-contract-completes-its-military-pivot/
[38] https://www.wired.com/story/elon-musk-lieutenant-gsa-ai-agency/
[39] https://www.usatoday.com/story/news/politics/2025/02/03/dems-elon-musk-doge-takeover-treasury/78187978007/
[40] https://www.techpolicy.press/anatomy-of-an-ai-coup/
[41] https://therevolvingdoorproject.org/tracking-the-doge-treasury-raid/
[42] http://debcha.org/
[43] https://www.penguinrandomhouse.com/books/612711/how-infrastructure-works-by-deb-chachra/
[44] https://www.wired.com/story/doge-tts-fired/
[45] https://www.wired.com/tag/elon-musk/
[46] https://labornotes.org/
[47] https://labornotes.org/2025/02/federal-workers-organize-against-billionaire-power-grab
[48] https://www.instagram.com/workingfamilies/p/DGLZz2CP9bH/
[49] https://emilyinyourphone.substack.com/p/everything-you-need-to-know-about
[50] https://bsky.app/profile/prisonculture.bsky.social
[51] https://docs.google.com/document/d/1OSWxykA1WHOi0vTPLAJDaCeVhR3uSfh7PhlCj4t4yT0/edit?tab=t.0
[52] https://ethanmarcotte.com/wrote/leaving-18f/#fnref-fn-margins
[53] https://www.npr.org/2025/02/12/nx-s1-5293382/x-elon-musk-doge-cfpb
[54] https://www.levernews.com/musk-just-scored-more-government-cash-while-pushing-education-cuts/
[55] https://ethanmarcotte.com/wrote/leaving-18f/#fnref-fn-conflicts
[56] https://ethanmarcotte.com/wrote/leaving-18f/#fnref-fn-datalake
[57] https://ethanmarcotte.com/wrote/tag/work
[58] https://ethanmarcotte.com/wrote/tag/jobs
[59] https://ethanmarcotte.com/wrote/tag/politics
[60] https://ethanmarcotte.com/wrote/tag/us-politics
[61] https://ethanmarcotte.com/wrote/tag/employment
[62] https://ethanmarcotte.com/wrote/tag/government
[63] https://ethanmarcotte.com/wrote/on-context/
[64] https://ethanmarcotte.com/wrote/the-bricks-we-lay/
[65] https://ethanmarcotte.com/wrote/free-faster/
[66] https://ethanmarcotte.com/wrote/hello-editorially/
[67] https://ethanmarcotte.com/wrote/
[68] https://ethanmarcotte.com/wrote/blog-questions-challenge/
[69] https://ethanmarcotte.com/
[70] https://ethanmarcotte.com/work/
[71] https://ethanmarcotte.com/books/you-deserve-a-tech-union/
[72] https://ethanmarcotte.com/books/you-deserve-a-tech-union/
[76] https://ethanmarcotte.com/wrote/leaving-18f/#
[77] https://ethanmarcotte.com/
[78] https://ethanmarcotte.com/wrote/
[79] https://ethanmarcotte.com/books/
[80] https://ethanmarcotte.com/work/
[81] https://ethanmarcotte.com/contact/
[82] https://follow.ethanmarcotte.com/@beep
[83] https://bsky.app/profile/ethanmarcotte.com
[84] https://www.linkedin.com/in/ethan-marcotte/
[85] https://ethanmarcotte.com/accessibility/
[86] https://ethanmarcotte.com/wrote/feed.xml
[87] https://ethanmarcotte.com/wrote/leaving-18f/#top
[88] https://ethanmarcotte.com/wrote/leaving-18f/#content

555
static/archive/muffinman-io-lxkzuw.txt Normal file
View File

@@ -0,0 +1,555 @@
[1]Jump to content
[2] [4] Art /[5] Blog /[6] Projects /[7] About /[8] Stats /Search
The Tiny Book of Great Joys
10. February 2025
Posted in [12]Random
· 19 minutes read
If you are interested in how I over-engineered the process of making a tiny
book for my wife, using AI, a pen plotter, a 3D printer, and a lot of time, you
are in the right place. The book is titled The Tiny Book of Great Joys (Mala
Knjiga Velikih Radosti in Serbian) , and here is how it turned out:
The Tiny Book of Great Joys sitting on the table slightly open with the title
page showing
[13] Photo 1 [14] Photo 1 [15] Photo 1
My wife is delighted with it, so it was worth all the effort.
This post will take you through the process. It will be a long one, but please
stick around - I promise there will be a lot of pretty pictures.
Here is the outline of the post:
• [16]The idea
• [17]Drawings
• [18]Text
• [19]Plotting
• [20]Bookbinding
• [21]The finished book
• [22]Timeline
• [23]Conclusion
The idea[24] #
I had this idea for a while after seeing something similar somewhere on the
internet.. Since then, I always wanted to make one for my wife - a physically
small book with a bunch of small drawings of our memories together, inside
jokes, and little things she likes.
I wanted the illustrations to be hand-drawn, and I had a plan to ask my friend
to do them. But I knew he would refuse any kind of payment, so I felt bad
adding more work to his plate. So I shelved the idea, but every now and then,
it would pop up in my head.
Fast forward a few years - we got a kid, and our routine completely changed. We
are enjoying it a lot, but it can be very exhausting, and every day seems
identical to the last. That's why I decided I needed to do something for her to
break the routine. The book idea seemed perfect - personal and handcrafted - so
I gave it a try.
To be able to do everything myself, I went to create digital drawings and then
draw them on paper using my trusty pen plotter.
With the idea in place, I moved on to creating the drawings - which turned out
to be a challenge of its own.
Drawings[25] #
For pen plotting, one needs vector files, so I started drawing in Figma.
Unfortunately, I quickly realized that my drawing skills would not get me the
result I had envisioned. Determined to do it this time, I decided to try using
AI to generate images.
Midjourney[26] #
I got myself a Midjourney subscription and started playing with it. It took a
lot of failed attempts to figure out how to get drawings that were simple and
had a strong hand-drawn feel to them. Even then, I ended up editing every one
of them, but more on that later.
One of the first images I was satisfied with (it didn't end up in the book,
though):
Black and white drawing of a cute fox curled up sleeping.
It took a lot of time, but it was fun. Failed attempts were often quirky and
funny, and I was learning how to use the tool. And it made me feel like a
secret agent, doing it next to my wife, who had no idea what I was up to.
A bunch of attempts at drawing Link from Zelda
Prompting[27] #
I may be wrong, but I think Midjourney wasn't built for the kind of
illustrations I had in mind. I was after simple, hand-drawn illustrations that
felt personal. Luckily, I found a style reference (--sref 230156437) that
worked well for my case. I used it to generate almost all of the drawings that
ended up in the book. For those who haven't used Midjourney - you can use
images as style references to influence the style of images you want to
generate.
Most of my images were generated using that sref code and a style weight (A
number that tells Midjourney how much the reference should influence the final
output) between 150 and 400 (it can go from 0 to 1000).
As for the prompts, these are the key terms I combined with the description and
the style reference:
• black and white
• vector line art
• stylized simple drawing
• solid white background
• isolated on white background
• low detail
• clean edges
• sketch
• rough sketch
• children's coloring book
It took me a lot of tries - between 10 and 30 attempts for each image you see
in the book.
AI to Plotter[28] #
Once I solved the image generation part, I had to figure out how to turn them
into vector files for plotting. The first thing I tried was something similar
to halftone. As you can see below, in this process, the images completely lost
the hand-drawn feel.
The same fox drawing, but but drawn with a lot of small dots, using technique
similar to halftone
Then I remembered [29]this plot of Marble Machine X I did a while ago, for
which I used AutoTrace to convert the original image to a vector file. The
great thing about AutoTrace is that it supports "centerline tracing". And this
time, I learned that Inkscape has a great AutoTrace plugin, which made it even
easier to convert.
What makes centerline tracing different[30] #
Most of the tools that convert raster to vector images do it by outlining
shapes. This is not suitable for plotting, as each line in the original image
becomes a sausage-like shape. Centerline tracing, on the other hand, tries to
draw a single line following the middle path through shapes. Don't worry if it
sounds confusing; the example below should make things clearer.
Here is the image of Link from The Legend of Zelda generated by Midjourney:
Black and white cartoony drawing of Link from The Legend of Zelda standing with
a sword and shield.
After applying a common vectorization technique, we get this. As you can see,
each line in the original drawing is now outlined, creating this messy-looking
image.
Vectorized image of Link using common vectorization technique with each area
outlined black
But if we use centerline tracing, it suddenly looks a lot more like a drawing.
It is not perfect, but don't worry - we are going to clean it up in the next
step.
Vectorized image of Link using centerline tracing, looking much more like a
real drawing
Cleaning up[31] #
In the points where lines touch or cross, AutoTrace is not sure which line to
follow and creates these funky-looking joints. Here is an exaggerated example
to show you what I'm talking about. Input is the raster image at the top and
the vectorized result is at the bottom:
Lines that are crossing and touching before and after centerline tracing
But I found out that if I roughly separate these lines, I get a much better
result.
Lines that are crossing and touching, but slightly separated before tracing,
with the result being much better
Let's now apply this technique to the image of Link we've seen above. After
separating lines (and some cleaning up) this is the image I ended up with. It
is rough, but it is only used as an input for the tracing process, so it
doesn't really matter. This was manual and somewhat tedious process, but I
enjoyed it overall. It was a sort of meditation for me.
The image of Link, but this time with lines slightly separated and details
removed
And finally, when we trace this image, we get a really nice and clean vector
file perfect for plotting. Very clean vectorized image of Link
Here is another example. We start with the image I generated using Midjourney:
Black and white drawing of a woman, man, little girl and a dog walking in a
forest
After editing, removing details and separating lines, we get this one:
The same image of the family walking in the forest but with lines separated and
some parts redrawn
And the traced vector result:
Vectorized image of of the family walking in the forest
You'll notice that in both examples I did some redrawing (For example, in the
second image, I redraw the dog completely to look like our dog Zappa.) . I did
that for pretty much all of the images, to fix things I wasn't able to polish
using prompts. I also removed a lot of details to make sure images are crisp
and readable at the small size.
Final image flow[32] #
All of this took a lot of experimentation, but it gave me a pretty solid
workflow which I used to generate all of the images. The complete flow looks
like this:
• Generate images using Midjourney.
• Upscale them two times, because upscaled images were easier to edit and
tracing was more precise.
• Clean up, redraw and separate lines by hand using Gimp.
• Use Inkscape plugin to run AutoTrace centerline tracing.
It took me a while to generate all the images, and the fact that I was trying
to keep it a secret from my wife didn't help. I think I did it over the span of
two weeks, mostly in the evening after she would go to bed.
Ganon (Name of the main villain Link fights against in The Legend of Zelda
series) never stood a chance![33] #
Before we continue I just want to show you two funky images of Link that really
made me laugh:
Funky looking Links generated by Midjourney
Midjourney please staph!
Text[34] #
With the drawings ready, I turned to the next crucial part - the text. I first
wanted to write everything by hand, photograph it and then vectorize it in the
same way I did with the images. But it was a hassle - I had to do a lot of
editing for text to look as my handwriting.
Evil Mad Scientist, the maker of my pen plotter, has a fantastic tool called
[35]Hershey Text. It contains a bunch of single-line fonts ideal for plotting.
I chose the EMS Elfin font as it looked playful and hand-drawn. I used it to
write all of the text in the book and I think it turned out great.
Here is how it looks:
Title of the book in English and Aerbian in EMS Elfin font
Plotting[36] #
The tricky part with bookbinding is that pages are not printed in order, but in
a way that when you fold the sheets in half, you get the right order. I used
Figma to design the layout, with a great care to make sure pages are in order
after double-sided plotting.
Here is the layout laid out on A4 sized paper. Sorry for blurring the text, but
a lot of it is very personal and I want to keep it for our eyes only.
Layout of the book ready for plotting
Plotting is the part that went the smoothest, but not without hiccups. I
usually use Pigma Micron blackliner markers. They use archival quality ink and
they are literally indestructible. But this time, even the thinnest one I had
was too thick for the book this small.
Here you can see the first two test plots (Sorry for the poor quality photos, I
threw the plots away, so these are the only ones I have) using markers of 0.2mm
and 0.1mm thickness respectively. Lines got a bit smudged and looked much
thicker than I expected. This was also the moment I realized I need to remove a
lot of details (A friend of mine said that in these plots, Link looks like he
has measles) from the images to make them readable at this size.
Test plot using 0.2mm marker Test plot using 0.1mm marker
I needed to find a thinner pen.
Technical pen to the rescue[37] #
Blackliner markers were made as a more practical replacement for technical
pens. But from what I've read, an old-school technical pen was the only thing
capable of achieving super-fine lines I wanted. I went online and ordered
Rotring Isograph 0.2mm. As soon as it arrived I sneaked out to my study and did
another test plot using it. Oh boy, was I happy when I saw the result:
Test plot using a technical pen
Lines were thin and crisp and at this point I was convinced the project will be
a success!
Smudged drawings[38] #
All of the first plots were done on 120gsm printer paper. It is somewhat thick
paper and drawings looked fantastic. Unfortunately, when I bound the pages
together, the drawings and letters would get transferred on the opposite pages.
I could probably get away with it, considering the whole hand made feel of the
book. But I wanted it to be perfect.
A friend advised me to leave ink to dry for a few hours. I left each side to
dry for 24 hours, but it smudged again. Next time I tried putting the plot
(before cutting the pages) between two sheets of papers and pressing it with
heavy books. I did that for more than 24 hours, but still after cutting and
bounding the pages, they got smudged again. At this point I was becoming
somewhat desperate. As the last resort I ordered different, 100gsm paper and to
my relief it worked! Crisis averted!
In the final version you can still see tiny traces on a few pages, but these
are barely visible and don't really bother me.
After plotting and cutting I was left with a stack of somewhat delicate pages.
Now, it was finally time to turn them into a book.
Bookbinding[39] #
As you can imagine, I had zero bookbinding experience. There are a lot of
resources online, but two of them were crucial for my project as they were on
how to bind tiny books:
• [40]Mini BookBinding Marathon video
• [41]How to Make A Miniature Hardback Book article
After reading and watching these and a few generic articles on bookbinding, I
gathered enough info to try doing it myself. I thought I was super clever
because I 3D printed sides and spine of the book. I designed sewing holes in
the spine so I can connect the pages directly to it without using glue. It was
a decent idea, but it left a gap between two signatures (In bookbinding, a
section, gathering, or signature is a group of sheets folded in half.) . Still,
I went with it for the first try.
3D printed spine with two book signatures already sewn to it
I laid everything down on the canvas that the book would be wrapped in and
started assembling it. But I made a crucial mistake - I used super glue. It
dries quickly, it is stiff, and doesn't glue 3D printed plastic well and it
dissolved the paper I used. Long story short, I made a mess. But I didn't
stress too much, I just proclaimed that version is a prototype and used it as a
learning experience.
I ordered proper bookbinding glue (PVA). While I was waiting for it, I focused
on properly sewing the pages together.
Sewing the pages[42] #
The first time I sewed the pages together, I poked the holes by hand and they
were somewhat uneven. Again, it was nothing major, but I didn't like it. So I
designed and 3D printed a simple tool to help me drill the holes evenly.
The tool has two parts, and the pages fit snugly between them. Both top and
bottom parts have holes, so I was able to put the needle through and poke
perfectly even holes in the pages. I'm very proud of this silly contraption.
3D printed tool, closed, with needle poking through it
3D printed tool, opened, with the sheet with sewing holes visible still in it
Here you can see all of the eight sheets with sewing holes.
All eight sheets with illustrations ready for sewing
Fun fact, I designed all 3D parts using JavaScript and [43]Replicad library.
Here is [44]a link if you want to play with the model in your browser.
[45]Application showing code and the 3D model
But I ditched the 3D printed spine and used the technique called pamphlet
stitch, which works great when you have only two signatures. It made signatures
way more tight than when I connected them separately to the 3D printed spine.
Two book signatures sewn together using pamphlet stich
Two book signatures opened at the exact point where they meet
Glue arrived[46] #
When the glue arrived, I plotted everything again and took it from the top. I
swapped 3D printed sides for cardboard. Using proper glue was a game changer. I
had enough time to apply it before it hardened, and when it dried it stayed
flexible. And when it got onto my fingers, it was easy to remove. Everything
was much cleaner, and I finally managed to put it all together.
Unfortunately, I was rushing to finish the book, so I didn't take any photos of
the process. But here are a few I do have:
Cardboard sides laid out in the bookbinding canvas
If you are an experienced bookbinder and reading this, I'm sorry for the
bookbinding crimes I probably committed. I promise I won't use super glue
again.
The finished book[47] #
It looked great! It was not perfect (more on that below), but I was super happy
with how it turned out. It had a distinct handcrafted feel to it, the images
turned out fantastic, and I think I really managed to bring out a personal
touch with it.
On the day I finished the book and gave it to my wife, we were both exhausted
(our kid was teething, and we had a very rough night), so I thought she would
appreciate a little pick-me-up.
Book opened on the table showing the illustration of the family walking in the
forrest
When I gave it to her, the first thing she asked was, "Will I cry?". She was
brave, but it definitely got her all mushy and made her day. After reading, she
carefully put it on the shelf, out of the reach of the little one.
Then I asked her if she ever suspected I was preparing a surprise for her, and
she said that she had no idea. But she also said that she thought it was weird
that I would often plot something and not brag about it to her afterwards. It
was true, I love showing her my work, but luckily she didn't give it too much
thought, and I was able to finish my secret project.
Book opened on the table showing the illustration of Link from The Legend of
Zelda
One thing I would like to fix[48] #
Like I mentioned, the book isn't perfect. The sides are a bit too large, so the
pages seem too deep inside when the book is closed. For the same reason, the
end pages turned out to be a bit short, which gives it a weird, uneven look. It
is purely aesthetic, but I think it is the only thing keeping it from being
perfect.
Lesson learned if I ever end up doing something similar.
Timeline[49] #
It took way longer than it should have—it took me a month and a half to finish
it. It took so long because I did it in secrecy, which meant working late in
the evenings when my wife and kid were asleep. A bunch of little failures...
ehm, I mean learning opportunities also prolonged the project. And finally, I
had to order multiple things, so I was blocked a few times while I was waiting
for four different deliveries.
But the final assembly took me around two and a half hours from start to finish
- plotting, cutting, sewing, and bookbinding. Mostly because I had already
practiced all of them and defined the exact process.
Conclusion[50] #
It was so much fun. I love projects that span across multiple disciplines. This
one touched AI, drawing, plotting, modeling, 3D printing, sewing, and
bookbinding. I encountered a lot of little hiccups, but I also learned about
all of them. Some of the errors I made could have been avoided if I had been
more patient. But I hope you'll cut me some slack - I was super excited and
eager to see how it would turn out, and I had limited time windows when I could
do it in secrecy. Still, I need to take it as a lesson - being patient will
help me save time when doing projects like this one.
The highlight for me was that I could do it without an illustrator. Love it or
hate it, AI ended up being a fantastic tool that filled the gap in my skill
set, which was crucial for making the book.
I hope you enjoyed this write-up as much as I enjoyed making the book and
writing the post. And I do hope I inspired you to try making something of your
own. If I did, please reach out on GitHub, I would love to see it.
Share on:
[51] [52] [53] [54]
Related Posts
[55]
Letters from Sarajevo
26. March 2020
[56]
Bunny jumps again
23. February 2025
[57]
Custom giraffe caret
20. June 2023
[58] RSS Feed
[59]
© 2016-2025. All rights reserved.
Written with ♡ by Stanko Tadić.
[60]Home/[61] Art /[62] Blog /[63] Projects /[64] About /[65] Stats /Search
Menu and search
[67]
Search[69][ ]
[70] Art [71] Blog [72] Projects [73] About [74] Stats [75] RSS Feed [76]
GitHub
References:
[1] https://muffinman.io/blog/the-tiny-book-of-great-joys/#content
[2] https://muffinman.io/
[4] https://muffinman.io/art/
[5] https://muffinman.io/blog/
[6] https://muffinman.io/projects/
[7] https://muffinman.io/about/
[8] https://muffinman.io/stats/
[12] https://muffinman.io/archive/#random
[13] https://muffinman.io/img/tiny-book/the-book-02.jpg
[14] https://muffinman.io/img/tiny-book/the-book-03.jpg
[15] https://muffinman.io/img/tiny-book/the-book-04.jpg
[16] https://muffinman.io/blog/the-tiny-book-of-great-joys/#the-idea
[17] https://muffinman.io/blog/the-tiny-book-of-great-joys/#drawings
[18] https://muffinman.io/blog/the-tiny-book-of-great-joys/#text
[19] https://muffinman.io/blog/the-tiny-book-of-great-joys/#plotting
[20] https://muffinman.io/blog/the-tiny-book-of-great-joys/#bookbinding
[21] https://muffinman.io/blog/the-tiny-book-of-great-joys/#the-finished-book
[22] https://muffinman.io/blog/the-tiny-book-of-great-joys/#timeline
[23] https://muffinman.io/blog/the-tiny-book-of-great-joys/#conclusion
[24] https://muffinman.io/blog/the-tiny-book-of-great-joys/#the-idea
[25] https://muffinman.io/blog/the-tiny-book-of-great-joys/#drawings
[26] https://muffinman.io/blog/the-tiny-book-of-great-joys/#midjourney
[27] https://muffinman.io/blog/the-tiny-book-of-great-joys/#prompting
[28] https://muffinman.io/blog/the-tiny-book-of-great-joys/#ai-to-plotter
[29] https://www.instagram.com/p/CNJ_ZBOHZKj/
[30] https://muffinman.io/blog/the-tiny-book-of-great-joys/#what-makes-centerline-tracing-different
[31] https://muffinman.io/blog/the-tiny-book-of-great-joys/#cleaning-up
[32] https://muffinman.io/blog/the-tiny-book-of-great-joys/#final-image-flow
[33] https://muffinman.io/blog/the-tiny-book-of-great-joys/#never-stood-a-chance
[34] https://muffinman.io/blog/the-tiny-book-of-great-joys/#text
[35] https://wiki.evilmadscientist.com/Hershey_Text
[36] https://muffinman.io/blog/the-tiny-book-of-great-joys/#plotting
[37] https://muffinman.io/blog/the-tiny-book-of-great-joys/#technical-pen-to-the-rescue
[38] https://muffinman.io/blog/the-tiny-book-of-great-joys/#smudged-drawings
[39] https://muffinman.io/blog/the-tiny-book-of-great-joys/#bookbinding
[40] https://www.youtube.com/watch?v=kA2bjvOAzGw
[41] https://www.rokolee.com/diy-miniature-hardback-book
[42] https://muffinman.io/blog/the-tiny-book-of-great-joys/#sewing-the-pages
[43] https://replicad.xyz/
[44] https://studio.replicad.xyz/workbench?from-url=https://muffinman.io/img/tiny-book/model.js
[45] https://studio.replicad.xyz/workbench?from-url=https://muffinman.io/img/tiny-book/model.js
[46] https://muffinman.io/blog/the-tiny-book-of-great-joys/#glue-arrived
[47] https://muffinman.io/blog/the-tiny-book-of-great-joys/#the-finished-book
[48] https://muffinman.io/blog/the-tiny-book-of-great-joys/#one-thing-i-would-like-to-fix
[49] https://muffinman.io/blog/the-tiny-book-of-great-joys/#timeline
[50] https://muffinman.io/blog/the-tiny-book-of-great-joys/#conclusion
[51] http://news.ycombinator.com/submitlink?u=https%3A//muffinman.io/blog/the-tiny-book-of-great-joys/&t=The%20Tiny%20Book%20%20of%20Great%20Joys
[52] http://twitter.com/share?text=The%20Tiny%20Book%20%20of%20Great%20Joys&url=https%3A//muffinman.io/blog/the-tiny-book-of-great-joys/
[53] https://bsky.app/intent/compose?text=The%20Tiny%20Book%20%20of%20Great%20Joys%20https%3A//muffinman.io/blog/the-tiny-book-of-great-joys/
[54] http://www.reddit.com/submit?url=https%3A//muffinman.io/blog/the-tiny-book-of-great-joys/&title=The%20Tiny%20Book%20%20of%20Great%20Joys
[55] https://muffinman.io/blog/letters-from-sarajevo/
[56] https://muffinman.io/blog/bunny-jumps-again/
[57] https://muffinman.io/blog/custom-giraffe-caret/
[58] https://muffinman.io/atom.xml
[59] https://muffinman.io/
[60] https://muffinman.io/
[61] https://muffinman.io/art/
[62] https://muffinman.io/blog/
[63] https://muffinman.io/projects/
[64] https://muffinman.io/about/
[65] https://muffinman.io/stats/
[67] https://muffinman.io/
[70] https://muffinman.io/art/
[71] https://muffinman.io/blog/
[72] https://muffinman.io/projects/
[73] https://muffinman.io/about/
[74] https://muffinman.io/stats/
[75] https://muffinman.io/atom.xml
[76] https://github.com/stanko/

358
static/archive/nshipster-com-b3vpys.txt Normal file
View File

@@ -0,0 +1,358 @@
[1]
[2]Ollama
Written by [3]Mattt February 14^th, 2025
“Only Apple can do this” Variously attributed to Tim Cook
Apple introduced [4]Apple Intelligence at WWDC 2024. After waiting almost a
year for Apple to, in Craig Federighi’s words, “get it right”, its promise of
“AI for the rest of us” feels just as distant as ever.
Can we take a moment to appreciate the name? Apple Intelligence. AI. That’s
some S-tier semantic appropriation. On the level of jumping on “podcast” before
anyone knew what else to call that.
While we wait for Apple Intelligence to arrive on our devices, something
remarkable is already running on our Macs. Think of it as a locavore approach
to artificial intelligence: homegrown, sustainable, and available year-round.
This week on NSHipster, we’ll look at how you can use Ollama to run LLMs
locally on your Mac — both as an end-user and as a developer.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[5]What is Ollama?
Ollama is the easiest way to run large language models on your Mac. You can
think of it as “Docker for LLMs” - a way to pull, run, and manage AI models as
easily as containers.
Download Ollama with [6]Homebrew or directly from [7]their website. Then pull
and run [8]llama3.2 (2GB).
$ brew install --cask ollama
$ ollama run llama3.2
>>> Tell me a joke about Swift programming.
What's a Apple developer's favorite drink?
The Kool-Aid.
Under the hood, Ollama is powered by [9]llama.cpp. But where llama.cpp provides
the engine, Ollama gives you a vehicle you’d actually want to drive — handling
all the complexity of model management, optimization, and inference.
Similar to how Dockerfiles define container images, Ollama uses Modelfiles to
configure model behavior:
FROM mistral:latest
PARAMETER temperature 0.7
TEMPLATE """
You are a helpful assistant.
User:
Assistant: """
Ollama uses the [10]Open Container Initiative (OCI) standard to distribute
models. Each model is split into layers and described by a manifest, the same
approach used by Docker containers:
{
"mediaType": "application/vnd.oci.image.manifest.v1+json",
"config": {
"mediaType": "application/vnd.ollama.image.config.v1+json",
"digest": "sha256:..."
},
"layers": [
{
"mediaType": "application/vnd.ollama.image.layer.v1+json",
"digest": "sha256:...",
"size": 4019248935
}
]
}
Overall, Ollama’s approach is thoughtful and well-engineered. And best of all,
it just works.
[11]What’s the big deal about running models locally?
[12]Jevons paradox states that, as something becomes more efficient, we tend to
use more of it, not less.
Having AI on your own device changes everything. When computation becomes
essentially free, you start to see intelligence differently.
While frontier models like GPT-4 and Claude are undeniably miraculous, there’s
something to be said for the small miracle of running open models locally.
• Privacy: Your data never leaves your device. Essential for working with
sensitive information.
• Cost: Run 24/7 without usage meters ticking. No more rationing prompts like
’90s cell phone minutes. Just a fixed, up-front cost for unlimited
inference.
• Latency: No network round-trips means faster responses. Your /M\d Mac((Book
( Pro| Air)?)|Mini|Studio)/ can easily generate dozens of tokens per
second. (Try to keep up!)
• Control: No black-box [13]RLHF or censorship. The AI works for you, not the
other way around.
• Reliability: No outages or API quota limits. 100% uptime for your [14]
exocortex. Like having Wikipedia on a thumb drive.
[15]Building macOS Apps with Ollama
Ollama also exposes an [16]HTTP API on port 11431 ([17]leetspeak for llama 🦙).
This makes it easy to integrate with any programming language or tool.
To that end, we’ve created the [18]Ollama Swift package to help developers
integrate Ollama into their apps.
[19]Text Completions
The simplest way to use a language model is to generate text from a prompt:
import Ollama
let client = Client.default
let response = try await client.generate(
model: "llama3.2",
prompt: "Tell me a joke about Swift programming.",
options: ["temperature": 0.7]
)
print(response.response)
// How many Apple engineers does it take to document an API?
// None - that's what WWDC videos are for.
[20]Chat Completions
For more structured interactions, you can use the chat API to maintain a
conversation with multiple messages and different roles:
let initialResponse = try await client.chat(
model: "llama3.2",
messages: [
.system("You are a helpful assistant."),
.user("What city is Apple located in?")
]
)
print(initialResponse.message.content)
// Apple's headquarters, known as the Apple Park campus, is located in Cupertino, California.
// The company was originally founded in Los Altos, California, and later moved to Cupertino in 1997.
let followUp = try await client.chat(
model: "llama3.2",
messages: [
.system("You are a helpful assistant."),
.user("What city is Apple located in?"),
.assistant(initialResponse.message.content),
.user("Please summarize in a single word")
]
)
print(followUp.message.content)
// Cupertino
[21]Generating text embeddings
[22]Embeddings convert text into high-dimensional vectors that capture semantic
meaning. These vectors can be used to find similar content or perform semantic
search.
For example, if you wanted to find documents similar to a user’s query:
let documents: [String] = …
// Convert text into vectors we can compare for similarity
let embeddings = try await client.embeddings(
model: "nomic-embed-text",
texts: documents
)
/// Finds relevant documents
func findRelevantDocuments(
for query: String,
threshold: Float = 0.7, // cutoff for matching, tunable
limit: Int = 5
) async throws -> [String] {
// Get embedding for the query
let [queryEmbedding] = try await client.embeddings(
model: "llama3.2",
texts: [query]
)
// See: https://en.wikipedia.org/wiki/Cosine_similarity
func cosineSimilarity(_ a: [Float], _ b: [Float]) -> Float {
let dotProduct = zip(a, b).map(*).reduce(0, +)
let magnitude = { sqrt($0.map { $0 * $0 }.reduce(0, +)) }
return dotProduct / (magnitude(a) * magnitude(b))
}
// Find documents above similarity threshold
let rankedDocuments = zip(embeddings, documents)
.map { embedding, document in
(similarity: cosineSimilarity(embedding, queryEmbedding),
document: document)
}
.filter { $0.similarity >= threshold }
.sorted { $0.similarity > $1.similarity }
.prefix(limit)
return rankedDocuments.map(\.document)
}
For simple use cases, you can also use Apple’s [23]Natural Language framework
for text embeddings. They’re fast and don’t require additional dependencies.
import NaturalLanguage
let embedding = NLEmbedding.wordEmbedding(for: .english)
let vector = embedding?.vector(for: "swift")
[24]Building a RAG System
Embeddings really shine when combined with text generation in a RAG (Retrieval
Augmented Generation) workflow. Instead of asking the model to generate
information from its training data, we can ground its responses in our own
documents by:
1. Converting documents into embeddings
2. Finding relevant documents based on the query
3. Using those documents as context for generation
Here’s a simple example:
let query = "What were AAPL's earnings in Q3 2024?"
let relevantDocs = try await findRelevantDocuments(query: query)
let context = """
Use the following documents to answer the question.
If the answer isn't contained in the documents, say so.
Documents:
\(relevantDocs.joined(separator: "\n---\n"))
Question: \(query)
"""
let response = try await client.generate(
model: "llama3.2",
prompt: context
)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
To summarize: Different models have different capabilities.
• Models like [25]llama3.2 and [26]deepseek-r1 generate text.
□ Some text models have “base” or “instruct” variants, suitable for
fine-tuning or chat completion, respectively.
□ Some text models are tuned to support [27]tool use, which let them
perform more complex tasks and interact with the outside world.
• Models like [28]llama3.2-vision can take images along with text as inputs.
• Models like [29]nomic-embed-text create numerical vectors that capture
semantic meaning.
With Ollama, you get unlimited access to a wealth of these and many more
open-source language models.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
So, what can you build with all of this?
Here’s just one example:
[30]Nominate.app
[31]Nominate is a macOS app that uses Ollama to intelligently rename PDF files
based on their contents.
Like many of us striving for a paperless lifestyle, you might find yourself
scanning documents only to end up with cryptically-named PDFs like
Scan2025-02-03_123456.pdf. Nominate solves this by combining AI with
traditional NLP techniques to automatically generate descriptive filenames
based on document contents.
The app leverages several technologies we’ve discussed:
• Ollama’s API for content analysis via the ollama-swift package
• Apple’s PDFKit for OCR
• The Natural Language framework for text processing
• Foundation’s DateFormatter for parsing dates
Nominate performs all processing locally. Your documents never leave your
computer. This is a key advantage of running models locally versus using cloud
APIs.
[32]Looking Ahead
“The future is already here – it’s just not evenly distributed yet.”
William Gibson
Think about the timelines:
• Apple Intelligence was announced last year.
• Swift came out 10 years ago.
• SwiftUI 6 years ago.
If you wait for Apple to deliver on its promises, you’re going to miss out on
the most important technological shift in a generation.
The future is here today. You don’t have to wait. With Ollama, you can start
building the next generation of AI-powered apps right now.
NSMutableHipster
Questions? Corrections? [33]Issues and [34]pull requests are always welcome.
This article uses Swift version 6.0. Find status information for all articles
on the [35]status page.
Written by Mattt
[36]Mattt
[37]Mattt ([38]@mattt) is a writer and developer in Portland, Oregon.
🅭 🅯 🄏 NSHipster.com is released under a [39]Creative Commons BY-NC License.
References:
[1] https://nshipster.com/
[2] https://nshipster.com/ollama/
[3] https://nshipster.com/authors/mattt/
[4] https://www.apple.com/apple-intelligence/
[5] https://nshipster.com/ollama/#what-is-ollama
[6] https://brew.sh/
[7] https://ollama.com/download
[8] https://ollama.com/library/llama3.2
[9] https://github.com/ggerganov/llama.cpp
[10] https://opencontainers.org/
[11] https://nshipster.com/ollama/#whats-the-big-deal-about-running-models-locally
[12] https://en.wikipedia.org/wiki/Jevons_paradox
[13] https://knowyourmeme.com/photos/2546581-shoggoth-with-smiley-face-artificial-intelligence
[14] https://en.wiktionary.org/wiki/exocortex
[15] https://nshipster.com/ollama/#building-macos-apps-with-ollama
[16] https://github.com/ollama/ollama/blob/main/docs/api.md
[17] https://en.wikipedia.org/wiki/Leet
[18] https://github.com/mattt/ollama-swift
[19] https://nshipster.com/ollama/#text-completions
[20] https://nshipster.com/ollama/#chat-completions
[21] https://nshipster.com/ollama/#generating-text-embeddings
[22] https://en.wikipedia.org/wiki/Word_embedding
[23] https://developer.apple.com/documentation/naturallanguage/
[24] https://nshipster.com/ollama/#building-a-rag-system
[25] https://ollama.com/library/llama3.2
[26] https://ollama.com/library/deepseek-r1
[27] https://ollama.com/blog/tool-support
[28] https://ollama.com/library/llama3.2-vision
[29] https://ollama.com/library/nomic-embed-text
[30] https://nshipster.com/ollama/#nominateapp
[31] https://github.com/nshipster/nominate
[32] https://nshipster.com/ollama/#looking-ahead
[33] https://github.com/NSHipster/articles/issues
[34] https://github.com/NSHipster/articles/blob/master/2025-02-14-ollama.md
[35] https://nshipster.com/status/
[36] https://nshipster.com/authors/mattt/
[37] https://github.com/mattt
[38] https://twitter.com/mattt
[39] https://creativecommons.org/licenses/by-nc/4.0/

588
static/archive/prayash-io-vrlx4z.txt Normal file
View File

@@ -0,0 +1,588 @@
[1]
prayash.ioprayash.io
[2]home[3]journal[4]software[5]music[6]feed[7]about
A Taste of Vanlife
May 31, 2024
Camping has always been one of my favorite hobbies, but I've always wondered
what life might be like on four wheels. It's difficult to think about living on
the road indefinitely—and understandably so. I'm not really into the idea of
getting rid of all of my posessions and living out of a van, constantly on the
move. But surely, there's a way to try out that lifestyle without making such a
drastic commitment?
Years ago, my wife and I set a plan in motion: to save enough money to take a
year off from work, travel the world, and immerse ourselves in our personal
passions. This adventure seemed like the perfect starting point for that dream.
I've shared more about that thought process in [8]Reflections from Apple.
After some research, we decided to rent a fully equipped Sprinter van from a
private owner through Outdoorsy. It felt like the ideal middle ground—no
permanent commitment, but still a chance to experience van life firsthand.
Plus, we've been wanting to spend a lot more time playing outside lately. Our
goal was simple: to explore the iconic national parks of the western United
States, with a touch of the Canadian Rockies for good measure. Below is an
illustration of the route we took.
Vanlife Expedition Map
We took a few days to prepare ourselves, picked up our van, loaded it with
every supply we could think of, and set off on an epic roadtrip—one which we
knew we'll remember for the rest of our lives.
This post will be a photoset with words interspersed in-between. My hope is
that it will inspire you to go on an adventure like this of your own. I'll try
to share all the spots we camped at, the hikes we tackled, and some general
observations about life in a van. It'll be a fun one.
Joshua Tree
Our adventure started 8 hours south of the Bay Area in the desert. We arrived
late in the night at [9]Indian Cove Campground, which is just outside of Joshua
Tree National Park. The night sky was cluttered with stars, and temperatures
were perfect.
Mercedes Sprinter Van at JoshuaTree
Mercedes Sprinter Van at Joshua Tree atNight
We took our time and leisurely hiked around some popular trails in the park
like Arch Rock and Cholla Cactus Garden.
Pallu CactusPhotoshoot
Cholla Cactus Garden at JoshuaTree
Prayash next tocactus
Even in the blistering sun, the views were excellent. Joshua Tree's landscape
is so unique.
Pallavi staring at a JoshuaTree
Pallavi admiring a field ofrocks
Getting to cook lunch right next to these prime spots was such a fun
experience. Pallavi got a sneak peek of her dreams of owning her own food truck
and letting happy customers pick off food from the countertop.
Pallavi lunchsession
food truckcustomer
Alabama Hills
After a short 4 hour drive, we arrived at [10]Tuttle Creek Campground, an
incredible first-come-first-serve campground right along the Eastern Sierra. We
typically prefer first come first serve because reservations are always so
difficult to grab, but the anxiety of not knowing where exactly you'll spend
the night can be daunting. That risk, sometimes, is so worth it because it
leads you to some incredible campgrounds that you may not have run into
otherwise. We could not believe that our view was this good. It felt like a
postcard moment.
Mercedes Sprinter Van Alabama HillsCampground
As the sun started to descend behind the mountains, we prepped for dinner and
just enjoyed the view in front of us. As per usual, Pallavi cooked up a storm,
so we ate like royalty. Pictured is Nepali style stir-fried chicken and veggies
with pickled radish and beaten rice.
Dinner Alabama HillsCampground
Chureito Pagoda BlueHour
The next morning, we got up early to catch the sunrise at [11]Movie Road. There
were some road closures which prevented us from being able to drive the van
there, so we just stopped on the side of the road to get a good look at Mt.
Whitney and friends light up at dawn.
Chureito Pagoda BlueHour
Fushimi Inari-taishaTorii
Sake barrels at MeijiJingu
We headed back to the campsite and decided to spend the afternoon lounging
around, playing music, and relaxing. Then, we headed off to our next campsite
in Death Valley.
Death Valley
Things got real interesting and fun as we drove the van towards one of the most
remote campsites within Death Valley: [12]Eureka Dunes. The road to get to camp
was a grueling 30 miles of washboard roads. We drove the van at a soul-crushing
8 MPH because of how rough the roads were. At times, it felt the van was going
to topple because of how bumpy it got. All of the cabinetry and fixtures within
the van were rattling – not a good sign, but we were already too far in to turn
back.
It was a vast, desert landscape. Not a single soul in sight. No cell service.
No gas station or water source nearby. It is the most remote we've ever been,
and quite frankly, we were a little nervous at first. Pallavi patiently drove
us through this one while I was busy capturing the moment through my camera,
sometimes even running out of the van to get that long telephoto shot from far
away – all in an attempt to capture just how vast of a landscape it was, and
how little and alone we were.
Chureito Pagoda Blue Hour
Fushimi Inari-taisha Torii
Sake barrels at Meiji Jingu
The silence of the landscape helped us calm our nerves as we eventually made it
to the campsite. No one was there, and it was just a patch of dirt with a
handful of fire pits. We parked the van right in front of the formidable Eureka
Dunes, and I quickly ran out to snap a photo before we set off on the hike up
to the top of the dunes. It was peak golden hour time.
Chureito Pagoda BlueHour
The deception of the dunes quickly became obvious. The summit was miles farther
than we initially estimated with lots of false summits to lead you on. The
winds were blistering, and you could feel a billion particles of sand flying
into every inch of your body at piercing speeds. It was a rough hike, but there
was so much beauty to be taken in there. Perhaps the most aggresive blend of
peace and chaos I've experienced.
Fushimi Inari-taishaTorii
Sake barrels at MeijiJingu
Sake barrels at MeijiJingu
Chureito Pagoda BlueHour
We couldn't climb to the top because it was so windy and painful. Walking along
the ridge of the dune was difficult in such harsh winds. The sky dusked and it
was no longer to safe to continue hiking through sandbowls. We set back for
camp for a nice dinner. Content that we could enjoy the rest of the night in
solitude. A couple campers arrived later in the night so we didn't end up being
all alone at the end, which was a comforting thought.
The next morning, we decide to give the dunes another shot. Conditions looked
favorable wind-wise, but temperatures were already in the 90s well before the
sun reached its zenith. We trekked back up the ridgeline of the dune to get
another good look at the dunes that were hiding beyond. We didn't get to see it
under golden light like the evening before, but it was still epic.
Death Valley EurekaDunes
Death Valley EurekaDunes
Death Valley EurekaDunes
To get out of Eureka Dunes and onto the rest of the park, it was another
soul-crushing 40 miles of washboard. We made it out in tip-top shape many hours
later, and the road led us straight to pavement and onto [13]Ubehebe Crater.
The mid-day sun was blistering, but we ventured out of the confines of our air
conditioned van to get a good glimpse of it anyway.
Ubehebe Crater Mid-DaySun
Moving on. [14]Dante's View and [15]Zabriskie Point were straight out of
another planet. The polychrome mountains definitely give off Iceland vibes.
There's no comparison, of course. But it's cool that the US has some alien-like
landscapes too.
Fushimi Inari-taishaTorii
Fushimi Inari-taishaTorii
Sake barrels at MeijiJingu
We headed back for camp at [16]Mesquite Spring Campground around sunset. The
night sky was incredible, and the galactic center was clearly visible to the
naked eye. At 2am, we got out of the van to get a good look at the Milky Way.
Truly a core memory.
Fushimi Inari-taishaTorii
Sake barrels at MeijiJingu
The next morning, we did our usual business – emptying out the grey water tank,
filling up with fresh water, and emptying the septic tank. All necessary chores
for van survival! We made a quick stop at [17]Artist's Palette. Pallavi whipped
up a delicious bowl of noodles and eggs.
Fushimi Inari-taishaTorii
Sake barrels at MeijiJingu
We then headed off to the Utah-Arizona border, ending the evening at [18]White
House Campground.
Antelope Canyon
The following morning, a Navajo-led tour of Lower Antelope Canyon was on the
books. It was a fascinating lesson with geology and history intertwined.
Snapping photos in the canyon is a fun exercise for the discerning artist, as
you must carefully frame your shot in an ocean of contours.
Fushimi Inari-taishaTorii
Sake barrels at MeijiJingu
Fushimi Inari-taishaTorii
Sake barrels at MeijiJingu
It was too windy to kayak through the canyons, so for the rest of the
afternoon, we spent lounging at a [19]viewpoint above Lake Powell, where I
captured this 5-image panorama of the vista.
CanyonlandsPanorama
Pallavi spent time writing on her journal while I lost myself on the little
Martin guitar. Nothing is finer than playing music in nature.
Pallavi lounging in the van at Lake Powell
Prayash at Lake Powell
Fushimi Inari-taisha Torii
Sake barrels at Meiji Jingu
Once we had recharged our batteries, we decided to catch the sunset at
Horseshoe Bend. Being one of the most photographed spots in the world, it was
no surprise that this would be a fantastic spot to catch last light.
CanyonlandsPanorama
Once the sun dipped over the horizon, we headed out to Zion to camp for the
night.
Zion
We had a dispersed campsite that we loved camping at years prior, so we were
stoked to revisit the same spot right under [20]Smithsonian Butte.
Unfortunately, all of the BLM campsites in that area had been taken out of
commission due to excessive noise and trash left by visitors. The locals were
rightfully annoyed, so BLM decided to break down all the sites. This is why we
can't have nice things (╯°□°）╯︵ ┻━┻.
I was heartbroken, and stressed out that we now had to hunt for another
campsite at midnight when we're already tired. One of my favorite sites was no
more, but here's how it looked years ago when we camped there for my birthday.
I'll share a photo of that instead.
Prayash at LakePowell
Anyway, we decided to just grab a vacant lot at a nearby RV park because it was
too late to hunt around for a campsite. The next morning, we headed straight
for Angel's Landing.
Canyonlands Hike
Canyonlands CandlestickTower
Canyonlands Panorama
It wasn't as scary as the ultra wide-angle videos on YouTube make it seem. The
exposure definitely sends chills down your spine, but there were plenty of
guardrails and chains to hold onto the entire way. Views of the entire Zion
Canyon were fantastic and we enjoyed one of our favorite snacks, Kurkure, at
the top.
Canyonlands Panorama
Canyonlands Hike
Canyonlands Candlestick Tower
Canyonlands Hike
Canyonlands Candlestick Tower
Canyonlands Panorama
We hastily rushed over to Bryce Canyon to spend the next couple of nights
there.
Bryce Canyon
Catching the sunrise at Bryce Canyon is one of the most magical things I've
ever experienced. It's a small park, but undeniably unique because of its
hoodoos. How magical it would be to see these features in their snowcapped
state, but I'll settle for an orange burst this time. And yet again, we're
humbled by the beauty of this planet.
Canyonlands Hike
Canyonlands CandlestickTower
CanyonlandsPanorama
CanyonlandsPanorama
The serene morning gradually transitioned to a fabulous breakfast and coffee at
the van. Pallavi got her morning yoga session in while I brewed coffee.
Self-care can be done even during busy periods of travel, folks! We took the
day slowly with a couple day hikes. A long day of walking is much more
preferable than a long day of driving.
CanyonlandsHike
Canyonlands CandlestickTower
CanyonlandsPanorama
We hiked the Navajo Loop trail and decided to end the night with a little jam
session by the campfire — a small cover of Junkeri by Bipul Chettri. A
beautifully poignant Nepali folk song. [21]You can watch our cover video here.
Canyonlands Hike
Canyonlands
The epic Utah excursion continues through Canyonlands, where we stared at the
vast abyss of Utah's desert in all directions. Striking rock formations and
atmospheric haze above the canyons and ravines. Utah is a place like no other.
We parked the van and lounged all day.
CanyonlandsCanyon
Canyonlands VanView
As a result of lazily lounging around, it was a nice balance to spend the next
day hiking.
CanyonlandsHike
Canyonlands CandlestickTower
CanyonlandsPanorama
A visit to Canyonlands must always be accompanied by a visit to Arches.
Arches
Arches has some absolutely epic structures, and the shifting light of day
really lets you appreciate them.
CanyonlandsHike
Canyonlands CandlestickTower
Canyonlands Hike
Canyonlands CandlestickTower
A quick coffee sesh with a view and a short walk, then we were off to Wyoming.
Grand Teton
On the way to the Tetons, we stopped by a farm for some Wyoming skyr. [22]
Shumway Farms is run by a kind family man who had all of his children helping
him run the farm. Unbelievably rich and fatty yogurt. It was unforgettable.
CanyonlandsHike
Canyonlands Hike
Canyonlands CandlestickTower
We made our way through Jackson Hole by the night and landed at our next
campsite at [23]Gros Ventre Campground. After a nutritious stir-fry, we hit the
hay early that night.
My alarm went off at 4:50 AM. Sunrise is at 5:05, and we're 20 minutes away
from [24]Schwabacher Landing, a classic sunrise spot for watching the Tetons
light up. In a delirious haze, we sped down the highway, took the turnout into
the small parking lot and hiked towards the river. At 5:10 AM, the Tetons were
beautifully highlighted and side-lit from the rising sun.
Oregon Coast God's Thumb
Oregon Coast God's Thumb
Canyonlands Hike
I whipped out the coffee grinder the moment we got back to our van for our
morning fix.
Oregon Coast God'sThumb
Oregon Coast God'sThumb
On our way northbound, the Tetons stared at us relentlessly. So we parked and
decided to stare back from the comfort of our van bed while the winds raged
outside.
Oregon Coast God'sThumb
Yellowstone
Upon arriving in Yellowstone, we spent the night at [25]Madison Campground. We
visited some of the classics like [26]Old Faithful and [27]Grand Prismatic
Spring. There were some truly wonderful textures there to photograph.
Oregon Coast God's Thumb Trail
Oregon Coast God's Thumb Trail
Oregon Coast God's Thumb Trail
Oregon Coast God's Thumb Trail
Oregon Coast God's Thumb Trail
Oregon Coast God's Thumb Trail
Glacier
The drive through Montana was stunning. It's easy to see why Montana so often
ranks as one of America's most beautiful states. It felt like we were in
Switzerland at times. Lush, green, and filled with epic mountains. We arrived
at Lake Macdonald right in time for blue hour. It was cold and nippy, a much
welcome change after all of those desert excursions.
Oregon Coast God's ThumbTrail
Oregon Coast God's ThumbTrail
The [28]Avalanche Lake trail was quite nice. Glacier wasn't quite in-season
yet, so it was the only accessible trail nearby.
The mountain pass on Going-to-the-Sun road was closed which meant you couldn't
drive to the eastern side without driving back outside of the park and around.
So we did exactly that, spending the night at [29]Saint Mary Campground. We
cooked and basked under a moonlit mesa.
Oregon Coast God's ThumbTrail
Oregon Coast God's ThumbTrail
East Glacier is absolutely epic. The Many Glaciers area in particular was
awesome. We did the [30]Grinell Glacier hike, but the end of the trail was
closed. We ran into a moose and saw a baby bear rustling around the bushes. It
was still a lovely hike, and ended with a ton of rain.
Oregon Coast God's Thumb Trail
Oregon Coast God's Thumb Trail
Oregon Coast God's Thumb Trail
Banff
Scattered thunderstorms in Banff led to little hiking, but grand views. Most of
our time was spent lounging around at different viewpoints, admiring the jagged
peaks of the Canadian Rockies. We camped at [31]Lake Louise Campground,
spending a full day admiring the peaks from Lake Louise.
Astoria Town Center
Astoria TownCenter
Astoria TownCenter
Astoria TownCenter
The next day, we hopped over to [32]Two Jack Lakeside Campground, which put us
in close proximity to this excellent lounging [33]viewpoint of Tunnel Mountain
Road and a fantastic [34]picnic viewpoint for Lake Minnewanka.
Astoria Town Center
Astoria Town Center
Astoria Town Center
Astoria Town Center
A loooong drive southwest through Idaho and Washington, and we ended up on
the...
Oregon Coast
The ocean, finally! Now we drive straight along the Oregon coastline starting
in Astoria. We explored this quaint little coastal town, sampling some seafood
and even picked up a cool record at a local vinyl shop called [35]The Lonely
Crab.
Astoria Town Center
Astoria Record Shop The LonelyCrab
We hiked [36]Indian Beach Trail in Ecola State Park, which ended with the
perfect lunch with a view of the seastacks, followed by a banger sunset at
Cannon Beach.
Oregon Coast God's Thumb Trail
Oregon Coast God's Thumb Trail
Oregon Coast God's Thumb Trail
Oregon Coast God's Thumb
Oregon Coast God's Thumb
Further south, we hiked to [37]God's Thumb, which ends at a cliff that juts out
into the sea. We lucked out with a beautiful, clear sunset at a place which is
notoriously gloomy on most days.
Oregon Coast God's ThumbTrail
Oregon Coast God's ThumbTrail
There, we met a lovely couple who were celebrating their wedding anniversary at
their first date spot. They were kind enough to snap a photo of us, too.
Oregon Coast God'sThumb
We continued south, heading towards Crater, with plenty of stops to cook and
admire the views along the way. [38]Tokeetee Falls was a good one. We stuck
through an opening in the boardwalk to get down closer to the waterfall.
Oregon Coast God's ThumbTrail
Oregon Coast God's ThumbTrail
Oregon Coast God's ThumbTrail
Jedediah Smith Redwoods
Our final stop, the much revered Redwoods. Some of the oldest growth in the
world, so much peace to be found among these titans. [39]Grove of Titans is a
highly recommended trail here.
Astoria TownCenter
Astoria TownCenter
Astoria TownCenter
Astoria TownCenter
Fin
This 1-month long taste of vanlife has been one of the greatest adventures of
our lives. We learned so much about ourselves. Many long drives and hikes
filled with vibrant conversations—some hard and some easy. It gave us a chance
to truly unwind from the stress and responsibilites of daily life.
Each day followed a simple rhythm: get from point A to point B, cook the best
meal we could manage, hike the most rewarding trail, or find the coolest
viewpoint. Purpose was easy to find, and that simplicity was liberating. There
was no need to think about work, errands, or chores to fill our time. The goal
was clear, and it made life feel beautifully uncomplicated.
It's definitely something we see ourselves doing at another phase of our life,
perhaps in 20 years we send off our future children to college. We met many
older couples in their 50s and 60s who were doing something similar, and they
praised us for taking the leap to do such a thing at a younger phase of our
lives.
The idea of a "mini-retirement" is compelling. Society often pushes the
narrative that we must spend all our youth studying, working, and building a
career to "save up for the future." But it's worth questioning this assumption.
Life isn't just about preparing for what's ahead; it's also about creating joy
and meaning in the present. We owe it to ourselves to find a balance—planning
wisely for the future while embracing happiness in the here and now.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Prayash Thapa
Hello! Thank you so much for visiting my blog. I hope you enjoyed your time
here. Photosets like these allow me to share my photographs with more context
and depth, and I find them rewarding to create. I hope to do more of these in
the future. Let's also connect on [40]Instagram or [41]Twitter.
• [42]← Chomolungma: Into the Heart of Khumbu
• [43]Reflections from Apple →
References:
[1] https://prayash.io/
[2] https://prayash.io/
[3] https://prayash.io/journal
[4] https://prayash.io/software
[5] https://prayash.io/music
[6] https://prayash.io/feed
[7] https://prayash.io/about
[8] https://prayash.io/journal/reflections-from-apple
[9] https://maps.app.goo.gl/imAyTZ27Y6Tekze58
[10] https://maps.app.goo.gl/nZX5cTy3ta4t8S3H8
[11] https://maps.app.goo.gl/izMZHaGMP3zUdx5B8
[12] https://maps.app.goo.gl/J94JXF2eTX1NEgYH9
[13] https://maps.app.goo.gl/QRVaDZdh4aEELBVM7
[14] https://maps.app.goo.gl/gkctY2faceqoFcns6
[15] https://maps.app.goo.gl/5YGiJHC9wYSy5UCq5
[16] https://prayash.io/journal/vanlife
[17] https://maps.app.goo.gl/AG9tbFaQynFFj1pA6
[18] https://prayash.io/journal/vanlife
[19] https://maps.app.goo.gl/DqkTbp57gNc4DrSH7
[20] https://maps.app.goo.gl/wMBG2RqYgKg5QrpV6
[21] https://youtube.com/pallavibhusal/videos
[22] https://www.shumwayfarms.com/product-page/icelandic-skyr-jh
[23] https://maps.app.goo.gl/YB4iFtMRUCggnoPQ8
[24] https://maps.app.goo.gl/cpyRpqe2casQwwN37
[25] https://maps.app.goo.gl/NzJ26yfC8N63d9PM9
[26] https://maps.app.goo.gl/B6HumotVCqduUGmE8
[27] https://maps.app.goo.gl/XoJtLaAL1AUswfmJ7
[28] https://www.alltrails.com/trail/us/montana/avalanche-lake--6
[29] https://maps.app.goo.gl/t7q9XHTLHxZKwu8b7
[30] https://prayash.io/journal/vanlife
[31] https://maps.app.goo.gl/mJMu1rwecc5Hg98m9
[32] https://maps.app.goo.gl/h7tvN7qkUPs5aRei8
[33] https://maps.app.goo.gl/yanZqJyQMQ7ikND96
[34] https://maps.app.goo.gl/GfWQAPTARiGW5Pkn9
[35] https://maps.app.goo.gl/2oJkD95jHh1Bv1AE9
[36] https://www.alltrails.com/trail/us/oregon/ecola-state-park-to-indian-beach-trail
[37] https://www.alltrails.com/trail/us/oregon/gods-thumb-via-the-knoll
[38] https://maps.app.goo.gl/mvLYJv3rM3j2sHa47
[39] https://www.alltrails.com/trail/us/california/grove-of-titans
[40] https://instagram.com/prayash.io
[41] https://x.com/prayash_io
[42] https://prayash.io/journal/into-the-heart-of-khumbu
[43] https://prayash.io/journal/reflections-from-apple

357
static/archive/www-404media-co-3wvica.txt Normal file
View File

@@ -0,0 +1,357 @@
Account
• [1]Log in
• [2]Subscribe
Navigation
• [3]Home
• [4]About
• [5]RSS
• [6]Support/FAQ
• [7]Podcast
• [8]FOIA Forum Archive
• [9]Merch
• [10]Advertise
• [11]Thanks
• [12]Privacy
Follow us
[13]Twitter [14]Bluesky [15]Mastodon [16]Instagram [17]TikTok [18]Facebook [19]
RSS
[20]
[21]Sign in [22]Subscribe
• [24]About
• [25]RSS
• [26]Support/FAQ
• [27]Podcast
• [28]FOIA Forum Archive
• [29]Merch
• [30]Advertise
• [31]Thanks
• [32]Privacy
Advertisement
•
[33]Go ad free
[34]organizing
You Can’t Post Your Way Out of Fascism
[35] Janus Rose
· Feb 5, 2025 at 11:47 AM
Authoritarians and tech CEOs now share the same goal: to keep us locked in an
eternal doomscroll instead of organizing against them, Janus Rose writes.
You Can’t Post Your Way Out of Fascism Unsplash / Collage via 404 Media
If there’s one thing I’d hoped people had learned going into the next four
years of Donald Trump as president, it’s that spending lots of time online
posting about what people in power are saying and doing is not going to
accomplish anything. If anything, it’s exactly what they want.
Trump’s second presidential term has arrived amidst a new golden age for
internet grifters, propagandists, and bad-faith hucksters of all stripes. The
contours of this era of untruth have been flashing like neon signs for the past
decade, constantly enticing us to engage with its impenetrable nonsense.
Whether it’s gaslighting everyone who saw Elon Musk give two Nazi salutes [36]
during the inauguration or blaming the Los Angeles wildfires [37]on the racist
dog whistle of “DEI,” lies and absurdities now regularly flood our senses,
having long outpaced the media’s capacity to filter them.
Many of my journalist colleagues have attempted to beat back the tide under
banners like “fighting disinformation” and “accountability.” While these
efforts are admirable, the past few years have changed my own internal
calculus. Thinkers like[38] Jean-Paul Sartre and[39] Hannah Arendt warned us
that the point of this deluge is not to persuade, but to overwhelm and paralyze
our capacity to act. More recently, researchers have found that the viral
outrage disseminated on social media in response to these ridiculous claims
actually [40]reduces the effectiveness of collective action. The result is a
media environment that keeps us in a state of debilitating fear and anger,
endlessly reacting to our oppressors instead of organizing against them.
To that end, the age of corporate social media has been a roaring success.
“The reality is you are oxygenating the things these people are saying even as
you purport to debunk them,” Katherine Cross, a sociologist and author of Log
Off: Why Posting and Politics (Almost) Never Mix, told 404 Media. “Whether it’s
[New York Times columnist] Ross Douthat providing [41]a sane-washing gloss on
Trump’s mania or people on social media vehemently disagreeing and dunking on
it, they’re legitimizing it as part of the discourse.”
Cross’ book contains a meticulous catalog of social media sins which many
people who follow and care about current events are probably guilty of—myself
very much included. She documents how tech platforms encourage us, through
their design affordances, to post and seethe and doomscroll into the void,
always reacting and never acting.
But perhaps the greatest of these sins is convincing ourselves that posting is
a form of political activism, when it is at best a coping mechanism—an
individualist solution to problems that can only be solved by collective
action. This, says Cross, is the primary way tech platforms atomize and
alienate us, creating “a solipsism that says you are the main protagonist in a
sea of NPCs.”
“Everything on social media is designed to make you think like that,” said
Cross. “It’s all about you—your feed, your network, your friends.”
In the days since the inauguration, I’ve watched people on Bluesky and
Instagram fall into these same old traps. My timeline is full of reactive hot
takes and gotchas by people who still seem to think they can quote-dunk their
way out of fascism—or who know they can’t, but simply can’t resist taking the
bait. The media is more than willing to work up their appetites. Legacy news
outlets cynically chase clicks (and ad dollars) by disseminating whatever
sensational nonsense those in power are spewing.
"For most people, social media gives you this sense that unless you care
about everything, you care about nothing. You must try to swallow the world
while it’s on fire"
This in turn fuels yet another round of online outrage, edgy takes, and
screenshots exposing the “hypocrisy” of people who never cared about being seen
as hypocrites, because that’s not the point. Even violent fantasies about
putting billionaires to the guillotine are rendered inept in these online
spaces—just another pressure release valve to harmlessly dissipate our rage
instead of compelling ourselves to organize and act.
This is the opposite of what media, social or otherwise, is supposed to do. Of
course it’s important to stay informed, and journalists can still provide the
valuable information we need to take action. But this process has been
short-circuited by tech platforms and a media environment built around seeking
reaction for its own sake. Many Twitter refugees made a good choice in
migrating from Musk’s X to Bluesky, carving out a new online space that is
inhospitable to bigoted debate bros and time-wasting trolls. But in their
enemies’ absence, many of these Left-leaning posters have just reverted to
dunking on each other, preferring the catharsis of sectarian conflict over the
hard work of organizing.
Under this status quo, everything becomes a myopic contest of who can best
exploit peoples’ anxieties to command their attention and energy. If we don’t
learn how to extract ourselves from this loop, none of the information we gain
will manifest as tangible action—and the people in charge prefer it that way.
It’s no surprise that tech billionaires like Musk, Jeff Bezos, and Mark
Zuckerberg have rushed to kiss the ring of the twice-ascendent Trump. The
marriage of big tech and Trumpworld should make clear that Silicon Valley and
authoritarians share the same goal: to crush dissent by keeping their would-be
opponents spinning on an endless hamster wheel of reactive anger. And just like
in the classic 1983 thriller WarGames, the only winning move is not to play.
That can be a tough pill to swallow when the internet is our main window into
the world, and that world seems to be rapidly falling apart. We gaze into our
phone-portals, paralyzed by the trance of the doomscroll, reacting and swiping
from one news article and hot take to another. Authoritarians issue frightening
proclamations that may or may not be legally enforceable, seizing our attention
and energy and ensuring that the process will repeat, ad infinitum.
So what is the alternative? If we log off, what exactly are we supposed to do
instead? How are we supposed to get information without constantly raising our
antennae into the noxious cumulonimbus cloud of social media?
It isn’t quite as simple as “touch grass,” but it also sort of is.
Trusted information networks have existed since long before the internet and
mass media. These networks are in every town and city, and at their core are
real relationships between neighbors—not their online, parasocial simulacra.
Here in New York City, in the week since the inauguration, I’ve seen large
groups mobilize to[42] defend migrants from anticipated ICE raids and provide
warm food and winter clothes for the unhoused after[43] the city closed
shelters and abandoned people in sub-freezing temperatures. Similar efforts are
underway in Chicago, where [44]ICE reportedly arrested more than 100 people,
and in other cities where ICE has planned or attempted raids, with volunteers
assigned to keep watch over key locations where migrants are most vulnerable.
A few weeks earlier, residents created [45]ad-hoc mutual aid distros in Los
Angeles to provide food and essentials for those displaced by the wildfires.
The coordinated efforts gave Angelenos a lifeline during the crisis, cutting
through the [46]false claims spreading on social media about looting and
out-of-state [47]fire trucks being stopped for “emissions testing.” Many mutual
aid groups in Los Angeles have not just been helping people affected by the
fires but have also focused on distributing information about how to learn
about and resist ICE raids in Los Angeles. It is no surprise that some of the
[48]largest and most coordinated protests in the early days of Trump’s term
have happened in Los Angeles, where thousands of anti-ICE protesters shut down
the 101 highway and several streets in downtown Los Angeles Sunday.
Some of these efforts were coordinated online over Discord and secure messaging
apps, but all of them arose from existing networks of neighbors and community
organizers, some of whom have been organizing for decades.
“For most people, social media gives you this sense that unless you care about
everything, you care about nothing. You must try to swallow the world while
it’s on fire,” said Cross. “But [49]we didn’t evolve to be able to absorb this
much info. It makes you devalue the work you can do in your community.”
It’s not that social media is fundamentally evil or bereft of any good
qualities. Some of my best post-Twitter moments have been spent goofing around
with mutuals on Bluesky, or waxing romantic about the joys of human creativity
and art-making in an increasingly AI-infested world. But when it comes to
addressing the problems we face, no amount of posting or passive info
consumption is going to substitute the hard, unsexy work of organizing.
It’s a lesson the Extremely Online Left still hasn’t fully learned, failing
where its political enemies succeed. Reactionary right-wing groups like the
homophobic and transphobic [50]Moms for Liberty—which seeks to ban books from
LGBTQ and BIPOC authors under the guise of “parental rights”—have claimed
political victories by seizing power one public school board and small town at
a time. Other reactionaries have similarly managed to take their pet grievances
about diversity and wokeness to the national level by moving from online
outrage to on-the-ground community organizing.
You can d­iscourse and quote-dunk and fact-check until you’re blue in the face,
but at a certain point, you have to stop and decide what truth you believe in.
The internet has conditioned us to constantly seek new information, as if
becoming a sponge of bad news will eventually yield the final piece of a
puzzle. But there is also such a thing as having enough information. As the[51]
internet continues to enshittify, maybe what we really need is to start
trusting each other and our own collective sense of what is true and good.
We don’t need any more irony-poisoned hot takes or cathartic, irreverent snark.
We need to collectively decide what kind of world we actually do want, and what
we’re willing to do to achieve it.
Janus Rose is New York City-based journalist, educator and artist whose work
explores the impacts of A.I. and technology on activists and marginalized
communities. Previously a senior editor at VICE, she has been published in
digital and print outlets including e-Flux Journal, DAZED Magazine, The New
Yorker, and Al Jazeera.
More like this
[52] The Dude Whose Brain Turned to Glass
[53]The Dude Whose Brain Turned to Glass
He was hanging out in an ancient Roman port town 2,000 years ago, when
something struck him (a deadly volcanic eruption).
[54] Becky Ferreira Becky Ferreira
· Mar 1, 2025
[55] Behind the Blog: Stunt Blogging and the 'Fuck It' Paradigm
[56]Behind the Blog: Stunt Blogging and the 'Fuck It' Paradigm
This week, we discuss stunt blogging, Signal pains, and murderous Reels.
[57] Samantha Cole Samantha Cole ,
[58] Joseph Cox Joseph Cox ,
[59] Emanuel Maiberg Emanuel Maiberg
· Feb 28, 2025
[60] The Digital Packrat Manifesto
[61]The Digital Packrat Manifesto
DRM and big tech's war on ownership has led me to make my own media libraries,
and you should too.
[62] Janus Rose
· Feb 28, 2025
Advertisement
•
[63]Go ad free
Advertisement
•
[64]Go ad free
•
Hide
Unparalleled access to hidden worlds both online and IRL.
404 Media is a new independent media company founded by technology journalists
Jason Koebler, Emanuel Maiberg, Samantha Cole, and Joseph Cox.
• [65]About
• [66]RSS
• [67]Support/FAQ
• [68]Podcast
• [69]FOIA Forum Archive
• [70]Merch
• [71]Advertise
• [72]Thanks
• [73]Privacy
[74]Twitter [75]Bluesky [76]Mastodon [77]Instagram [78]TikTok [79]Facebook [80]
RSS
Join the newsletter to get the latest updates.
[81][ ]
Success
Great! Check your inbox and click the link to confirm your subscription
© 2025 404 Media. Published with [83]Ghost.
References:
[1] https://www.404media.co/signin/
[2] https://www.404media.co/signup/
[3] https://www.404media.co/
[4] https://www.404media.co/about/
[5] https://www.404media.co/404-media-now-has-a-full-text-rss-feed/
[6] https://www.404media.co/faq/
[7] https://www.404media.co/the-404-media-podcast/
[8] https://www.404media.co/foia-forum-archive/
[9] https://404media.myshopify.com/
[10] https://www.404media.co/advertise-with-404-media/
[11] https://www.404media.co/thank-yous/
[12] https://www.404media.co/privacy-policy/
[13] https://twitter.com/404mediaco
[14] https://bsky.app/profile/404media.co
[15] https://mastodon.social/@404mediaco
[16] https://instagram.com/404mediaco
[17] https://tiktok.com/@404.media
[18] https://www.facebook.com/404mediaco
[19] https://www.404media.co/rss
[20] https://www.404media.co/
[21] https://www.404media.co/signin/
[22] https://www.404media.co/signup/
[24] https://www.404media.co/about/
[25] https://www.404media.co/404-media-now-has-a-full-text-rss-feed/
[26] https://www.404media.co/faq/
[27] https://www.404media.co/the-404-media-podcast/
[28] https://www.404media.co/foia-forum-archive/
[29] https://404media.myshopify.com/
[30] https://www.404media.co/advertise-with-404-media/
[31] https://www.404media.co/thank-yous/
[32] https://www.404media.co/privacy-policy/
[33] https://www.404media.co/you-cant-post-your-way-out-of-fascism/#/portal/signup
[34] https://www.404media.co/tag/organizing/
[35] https://www.404media.co/author/janus/
[36] https://apnews.com/article/musk-gesture-salute-antisemitism-0070dae53c7a73397b104ae645877535?ref=404media.co
[37] https://www.politico.com/news/2025/01/15/republicans-dei-la-fires-00198551?ref=404media.co
[38] https://www.goodreads.com/quotes/7870768-never-believe-that-anti-semites-are-completely-unaware-of-the-absurdity?ref=404media.co
[39] https://www.openculture.com/2022/02/hannah-arendt-explains-how-propaganda-uses-lies.html?ref=404media.co
[40] https://par.nsf.gov/servlets/purl/10095997?ref=404media.co
[41] https://www.nytimes.com/2025/01/11/opinion/columnists/o-canada-come-join-us.html?ref=404media.co
[42] https://abc7ny.com/post/nyc-immigration-ice-agents-arrest-hundreds-migrants-sanctuary-cities-including-new-york/15835897/?ref=404media.co
[43] https://www.msn.com/en-us/news/us/nyc-to-close-13-more-migrant-shelters-by-june-2025/ar-BB1rfS3Y?ref=404media.co
[44] https://www.chicagotribune.com/2025/01/27/chicago-ice-raids-what-to-know/?ref=404media.co
[45] https://itsgoingdown.org/report-from-los-angeles-mutual-aid-hubs-fires/?ref=404media.co
[46] https://www.reuters.com/fact-check/video-shows-people-evacuating-home-not-looting-amid-la-fires-2025-01-13/?ref=404media.co
[47] https://www.thereflector.com/stories/no-out-of-state-fire-vehicles-turned-away-from-california-due-to-emissions-testing,373975?ref=404media.co
[48] https://abc7.com/post/dozens-march-101-freeway-downtown-los-angeles-during-protest-deportations/15857649/?ref=404media.co
[49] https://pmc.ncbi.nlm.nih.gov/articles/PMC6502424/?ref=404media.co
[50] https://www.vice.com/en/article/the-group-banning-lgbt-books-wants-to-replace-them-with-anti-gay-propaganda/?ref=404media.co
[51] https://gizmodo.com/enshittification-is-officially-the-biggest-word-of-the-year-2000530173?ref=404media.co
[52] https://www.404media.co/the-dude-whose-brain-turned-to-glass/
[53] https://www.404media.co/the-dude-whose-brain-turned-to-glass/
[54] https://www.404media.co/author/becky/
[55] https://www.404media.co/behind-the-blog-stunt-blogging-and-the-fuck-it-paradigm/
[56] https://www.404media.co/behind-the-blog-stunt-blogging-and-the-fuck-it-paradigm/
[57] https://www.404media.co/author/samantha-cole/
[58] https://www.404media.co/author/joseph-cox/
[59] https://www.404media.co/author/emanuel-maiberg/
[60] https://www.404media.co/the-digital-packrat-manifesto/
[61] https://www.404media.co/the-digital-packrat-manifesto/
[62] https://www.404media.co/author/janus/
[63] https://www.404media.co/you-cant-post-your-way-out-of-fascism/#/portal/signup
[64] https://www.404media.co/you-cant-post-your-way-out-of-fascism/#/portal/signup
[65] https://www.404media.co/about/
[66] https://www.404media.co/404-media-now-has-a-full-text-rss-feed/
[67] https://www.404media.co/faq/
[68] https://www.404media.co/the-404-media-podcast/
[69] https://www.404media.co/foia-forum-archive/
[70] https://404media.myshopify.com/
[71] https://www.404media.co/advertise-with-404-media/
[72] https://www.404media.co/thank-yous/
[73] https://www.404media.co/privacy-policy/
[74] https://twitter.com/404mediaco
[75] https://bsky.app/profile/404media.co
[76] https://mastodon.social/@404mediaco
[77] https://instagram.com/404mediaco
[78] https://tiktok.com/@404.media
[79] https://www.facebook.com/404mediaco
[80] https://www.404media.co/rss
[83] https://ghost.org/

40
static/archive/www-manton-org-iaxj45.txt Normal file
View File

@@ -0,0 +1,40 @@
● [1]Manton Reece
[2]About [3]Photos [4]Archive [5]30 days [6]88 parks [7]Replies [8]Reading [9]
Search [10]Also on Micro.blog
[11]Feb 9, 2025
I missed the memo for the outside space at this coffee shop… There are like 20
little kids and families running around here. Not a distraction, just makes me
smile remembering how good life was with little kids. The bittersweet irony
with parenting is not knowing until years later what you had.
[12]Also on Bluesky [13] [avatar]
Manton Reece [14]@manton
• [15]RSS
• [16]JSON Feed
• [17]Surprise me!
• [18]Tweets
References:
[1] https://www.manton.org/
[2] https://www.manton.org/about/
[3] https://www.manton.org/photos/
[4] https://www.manton.org/archive/
[5] https://www.manton.org/30-days/
[6] https://www.manton.org/88-parks/
[7] https://www.manton.org/replies/
[8] https://www.manton.org/reading/
[9] https://www.manton.org/search/
[10] https://micro.blog/manton
[11] https://www.manton.org/2025/02/09/i-missed-the-memo-for.html
[12] at://did:plc:pko7wbcggok753hnvndxh3ni/app.bsky.feed.post/3lhr2bc63zb2a
[13] https://www.manton.org/
[14] https://micro.blog/manton
[15] https://www.manton.org/feed.xml
[16] https://www.manton.org/feed.json
[17] https://www.manton.org/surprise-me/
[18] https://www.manton.org/tweets/

335
static/archive/www-robinsloan-com-dkgq9f.txt Normal file
View File

@@ -0,0 +1,335 @@
[1]Blog [2]About [3]Moonbound
This is a post from [4]Robin Sloan’s lab blog & notebook. You can [5]visit the
blog’s homepage, or [6]learn more about me.
[7]Is it okay?
February 11, 2025 Macbeth Consulting the Witches, 1825, Eugène Delacroix [8]
Macbeth Consulting the Witches, 1825, Eugène Delacroix
How do you make a language model? Goes like this: erect a trellis of code, then
allow the real program to grow, its development guided by a grueling training
process, fueled by reams of text, mostly scraped from the internet. Now. I want
to take a moment to think together about a question with no remaining practical
importance, but persistent moral urgency:
Is that okay?
The question doesn’t have any practical importance because the AI companies — 
and not only the companies, but the enthusiasts, all over the world — are going
to keep doing what they’re doing, no matter what.
The question does still have moral urgency because, at its heart, it’s a ques
tion about the things people all share together: the hows and the whys of
humanity’s common inheritance. There’s hardly anything bigger.
And, even if the companies and the enthusiasts rampage ahead, there are still
plenty of us who have to make personal decisions about this stuff every day.
You gotta take care of your own soul, and I’m writing this because I want to
clarify mine.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
A few ground rules.
First, if you (you engineer, you AI acolyte!) think the answer is obviously
“yes, it’s okay”, or if you (you journalist, you media executive!) think the
answer is obviously “no, it’s not okay”, then I will suggest that you are not
thinking with sufficient sensitivity and imagination about something truly new
on Earth. Nothing here is obvious.
Second, I’d like to proceed by depriving each side of its best weapon.
On the side of “yes, it’s okay”, I will insist that the analogy to human
learning is not admissible. “Don’t people read things, and learn from them, and
produce new work?” Yes, but speed and scale always influence our judgments
about safety and permissibility, and the speed and scale of machine learning is
off the charts. No human, no matter how well-read, could ever field requests
from a million other people, all at once, forever.
On the side of “no, it’s not okay”, I will set aside any arguments grounded in
copyright law. Not because they are irrelevant, but because … well, I think
modern copyright is flawed, so a victory on those grounds would be thin, a bit
sad. Instead, I’ll defer to deeper precedents: the intuitions and aspirations
that gave rise to copyright in the first place. To promote the Progress of Sci
ence and useful Arts, remember?
I hope partisans of both sides will agree this is a fair swap. Put down your
weapons, and let’s think together.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
I want to go carefully, step by step — yet I want to do so with brevity. Lan
guage models produce so … many … WORDS, and they seem to coax just as many out
of their critics. Logorrhea begets logorrhea. We can do better.
I’ll begin with my sense of what language models are doing. Here it is: lan
guage models collate and precipitate all the diverse reasons for writing,
across a huge swath of human activity and aspiration. Start to enumerate those
reasons: to inform, to persuade, to sell this stupid alarm clock, to dump the
CUSTOMERS table into a CSV file … and you realize it’s a vast field of desire
and action, impossible to hold in your head.
The language models have many heads.
In this formulation, language models are not merely trained on human writing.
They are the writing: all those reasons, granted the ability to speak for
themselves. I imagine the PyTorch code as a mech suit, with squishy language
strapped in tight … 
To make this work — you already know this, but I want to underscore it — only a
truly rich trove of writing suffices. Train a language model on all of
Shakespeare’s works and you won’t get anything useful, just a brittle
Shakespeare imitator.
In fact, the only trove known to produce noteworthy capabilities is: the entire
internet, or close enough. The whole extant commons of human writing. From here
on out, for brevity, we’ll call it Everything.
This is what makes these language models new: there has never, in human
history, been a way to operationalize Everything. There’s never been any
thing close.
Just as, above, I set copyright aside, I want also to set aside fair use and
the public domain. Again, not because they are irrelevant, but because those
intuitions and frameworks all assume we are talking about using some part of
the commons — not all of it.
I mean: ALL of it!
If language models worked like cartoon villains, slurping up Everything and
tainting it with techno-ooze, our judgment would be easy. But of course, digiti
zation is trickier than that: the airy touch of the copy complicates the sce
nario.
The language model reads Everything, and leaves Everything untouched — yet sud
denly this new thing exists, with strange and formidable powers.
Is that okay?
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
As we begin to feel our way across truly new terrain, we can inquire: how much
of the value of these models comes from Everything? If the fraction was just
one percent, or even ten, then we wouldn’t have much more to say.
But the fraction is, for sure, larger than that.
What goes into a language model? Data and compute.
For the foundation models like Claude, data means: Everything.
Compute combines two pursuits:
1. software: the trellises and applications that support the development and
deployment of these models, and
2. hardware: the vast sultry data centers, stocked with chips, that give them
room to run
There’s a lot of value in those pursuits; I don’t take either for granted, or
the labor they require. The experience you get using a model like Claude
depends on an ingenious scaffolding. [9]Truly! At the same time: I believe
anyone who works on these models has to concede that the trellises and the
chips, without data, are empty vessels. Inert.
Reasonable people can disagree about how the value breaks down. While I believe
the relative value of Everything in this mix is something close to 90%, I’m
willing to concede a 50/50 split.
And here is the important thing: there is no substitute.
You’ve probably heard about the race to generate novel training data, and all
the interesting effects such data can have. It is sometimes lost in those dis
cussions that these sophisticated new curricula can only be provided to a lan
guage model already trained on Everything. That training is what allows it to
make sense of the new material.
Also, it is often the case — not always, but often — that the novel training
data is generated by … a language model … which has itself been trained
on … you guessed it.
It’s Everything, all the way down.
Would it be possible to commission a fresh body of work, Everything’s equal in
scale and diversity, without any of the encumbrances of the commons? If you
could do it, and you trained a clean-room model on that writing alone, I con
cede that my question would be moot. (There would be other questions! Just not
this one.) Certainly, with as much money as the AI companies have now, you’d
expect they might try. We know they are already paying to produce new content,
lots of it, across all sorts of business and technical domains.
But this still wouldn’t match the depth and richness of Everything. I have a
hypothesis, which naturally might be wrong: that it is precisely the naivete of
Everything, the fact that its writing was actually produced for all those dif
ferent reasons, that makes it so valuable. Composing a fake corporate email,
knowing it will be used to train a language model, you’re not doing nothing,
but you’re not doing the same thing as the real email-writer. Your document
doesn’t have the same … what? The same grain. The same umami.
Maybe one of these companies will spend ten billion dollars to commission a
whole new internet’s worth of text and prove me wrong. However, I think there
are information-theoretic reasons to believe the results of such a project
would disappoint them.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
So! Understanding that these models are reliant on Everything, and derive a
large fraction of their value from it, one judgment becomes clear:
If their primary application is to produce writing and other media that crowds
out human composition, human production: no, it’s not okay.
For me, this is intuitively, almost viscerally, obvious. Here is the ultimate
act of pulling the ladder up behind you, a giant “fuck you” to every human who
ever wanted to accomplish anything, who matched desire to action, in writing,
part of Everything. Here is a technology founded in the commons, working to
undermine it. Immanuel Kant would like a word.
Fine. But what if that isn’t the primary application? What if language models,
by collating and precipitating all the diverse reasons for writing, become flex
ible general-purpose reasoners, and most of their “output” is never actually
read by anyone, instead running silent like the electricity in your walls?
It’s possible that language models could go on broadening and deepening in this
way, and eventually become valuable [10]aids to science and technology, [11]to
medicine and more.
This is tricky — it’s so, so tricky — because the claim is both (1) true, and
(2) convenient. One wishes it wasn’t so convenient. Can’t these companies
simply promise, with every passing year, that AI super science is just around
the corner … and meanwhile, wreck every creative industry, flood the internet
with garbage, grow rich on the value of Everything? Let us cook—while culture
fades into a sort of oatmeal sludge.
They can do that! They probably will. And the claim might still be true.
If super science is a possibility — if, say, Claude 13 can help deliver cures
to a host of diseases — then, you know what? Yes, it is okay, all of it. I’m
not sure what kind of person could insist that the maintenance of a media
status quo trumps the eradication of, say, most cancers. Couldn’t be me. Fine,
wreck the arts as we know them. We’ll invent new ones.
(I know that seems awfully consequentialist. Would I sacrifice anything, or
everything, for super science? No. But art and media can find new forms. That’s
what they do.)
Obviously, this scenario is especially appealing if the super science, like
Everything at its foundation, flows out into the commons. It should.
So — is super science really on the menu? We don’t have any way of knowing; not
yet. Things will be clearer in a few years, I think. There will either be real
undeniable glimmers, reported by scientists putting language models to work, or
there will still only be visions.
For my part, I think the chance of super science is below fifty percent, owing
mostly to the friction of the real physical world, which the language models
have, so far, avoided. But, I also think the chance is above ten percent, so,
I remain curious.
It’s not unreasonable to find this wager suspicious, but if you do, I might
ask: is there any possible-but-unproven technology that you think is worth pur
suing even at the cost of itchy uncertainty in the present? If the answer is
“yes, just not this one”: fair enough. If the answer is “no”: aha! I see you’ve
answered the question at the top of this page for yourself already.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Where does this leave us?
I suppose it’s not surprising, in the end:
If an AI application delivers some profound public good, or even if it might,
it’s probably okay that its value is rooted in this unprecedented operational
ization of the commons.
If an AI application simply replicates Everything, it’s probably not okay.
I’ll sketch out my current opinions more specifically:
I think the image generation models, trained on the Everything of pictures,
are: probably not okay. They don’t do anything except make more images. They
pee in the pool.
I think the foundation models like Claude are: probably okay. If it seemed, a
couple of years ago, that they were going to be used mainly to barf out text,
that impression has faded. It’s clear their applications are diverse, and often
have more to do with processes than end products.
The case of translation is compelling. If language models are, indeed, the
Babel fish, they might justify the operationalization of the commons even
without super science.
I think the case of code is especially clear, and, for me, basically settled.
That’s both (1) because of where code sits in the creative process, as an inter
mediate product, the thing that makes the thing, and (2) because the commons of
open-source code has carried the expectation of rich and surprising reuse for
decades. I think this application has, in fact, already passed the threshold of
“profound public good”: opening up programming to whole new groups of people.
But, again, it’s important to say: the code only works because of Everything.
Take that data away, train a model using GitHub alone, and you’ll get a far
less useful tool.
Maybe (it turns out) I’m less interested in litigating my foundational question
and more interested in simply insisting on the overwhelming, irreplaceable con
tribution of this great central treasure: all of us, writing, for every conceiv
able reason; desire and action, impossible to hold in your head.
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Did we make progress here? I think so. It’s possible my question, at the
outset, seemed broad. In fact, it’s fairly narrow, about this core mechanism,
the operationalization of the commons: whether I can live with it, or not.
One extreme: if these machines churn through all media, and then, in their
deployment, blow away any prospect for a healthy market for human-made media,
I’d say, no, that’s not what we want from technology, or from our future.
Another extreme: if these machines churn through all media, and then, in their
deployment, discover several superconductors and cure all cancers, I’d say,
okay … we’re good.
What if they do both? Well, it would be a bummer for media, but on balance I’d
take it. There will always be ways for artists to get out ahead again. More on
that in another post.
I also think there are some potential policy remedies that would even out the
allocation of value here — although, these days, imagining interesting policy
is a sort of fantastical entertainment. Even so, I’ll post about those later,
too.
In this discussion, I set copyright and fair use aside. I should say, however,
that I’m not at all interested in clearing the air for AI companies, legally.
They’ve chosen to plunge ahead into new terrain — so let them enjoy the fog of
war, Civ-style. Let them cook!
[12]To the blog home page
I'm [13]Robin Sloan, a fiction writer. The main thing to do here is sign up for
my newsletter:
[14][ ] [15][Subscribe]
This website doesn’t collect any information about you or your reading.
It aspires to the speed and privacy of the printed page.
Don’t miss [16]the colophon. Hony soyt qui mal pence
References:
[1] https://www.robinsloan.com/lab/
[2] https://www.robinsloan.com/about/
[3] https://www.robinsloan.com/moonbound/
[4] https://www.robinsloan.com/
[5] https://www.robinsloan.com/lab/
[6] https://www.robinsloan.com/about/
[7] https://www.robinsloan.com/lab/is-it-okay/
[8] https://www.clevelandart.org/art/1962.109?utm_source=Robin_Sloan_sent_me
[9] https://www.youtube.com/watch?v=ugvHCXCOmm4#t=9780
[10] https://research.google/blog/accelerating-scientific-breakthroughs-with-an-ai-co-scientist/?utm_source=Robin_Sloan_sent_me
[11] https://darioamodei.com/machines-of-loving-grace?utm_source=Robin_Sloan_sent_me
[12] https://www.robinsloan.com/lab/
[13] https://www.robinsloan.com/about?utm_source=Robin_Sloan_sent_me
[16] https://www.robinsloan.com/colophon/