davideisinger.com

My personal website
Log | Files | Refs | README

www-wrecka-ge-rdi1xr.txt (23459B)


      1 [1]wreckage/salvage
      2 [2][ ] 
      3 
      4   • [3]about
      5   • [4]posts
      6 
      7   • [5]sign up (free or paid)
      8   • [6]sign in
      9   • Search
     10 
     11 Bad shape
     12 21 days ago by [7]Erin Kissane — 11 min read
     13 
     14 Bad shape
     15 
     16 The idea I keep coming back to is that the big platforms, like Dickens' Marley,
     17 were dead to begin with, and are now something particularly bad, which is dead
     18 on their feet.
     19 
     20 In the game of Go, bad shape is the term for configurations of stones on the
     21 game board that are inefficient in achieving their offensive goal (territory
     22 capture) and unlikely to achieve their defensive goal (the state of "[8]life").
     23 You can extend a bad shape in a fruitless attempt to make it better, but you'll
     24 generally be wasting your time.
     25 
     26 The idea I keep coming back to is that the big platforms, [9]like Marley, were
     27 dead to begin with, and are now something particularly bad, which is dead on
     28 their feet. Not because they’re been abandoned by users (yet) but because
     29 they’re structurally incapable of governing the systems they made, and most of
     30 the things they try to do about it introduce more and weirder problems.
     31 
     32 While they were still gobbling hundreds of millions of new users each year—and
     33 while the old political machines were still catching up—platforms could outrun
     34 the problem. Now, though, the number of remaining uncaptured humans dwindles,
     35 the politicians and propagandists have adapted to exploit the mass-scale
     36 machinery the platforms provide, and the positions platform companies have
     37 contorted themselves into trying to shoehorn governance into ungovernable
     38 structures are increasingly hard to maintain.
     39 
     40 Facebook especially is likely to zombie-shamble along for some time, held
     41 upright by its deep weave into the coordination of offline life and by [10]the
     42 communities for whom it serves as a sole accessible connection to the internet,
     43 but the whole apparatus looks increasingly precarious.
     44 
     45 (These are very simple points, but it remains a wince-inducing faux pas to say
     46 them in a lot of tech-thinking spaces, so I will keep pushing on the obvious.)
     47 
     48 The evidence of the past decade and a half argues strongly that platform
     49 corporations are structurally incapable of good governance, primarily because
     50 most of their central aims (continuous growth, market dominance, profit via
     51 extraction) conflict with many basic human and societal needs.
     52 
     53 As entities, large social platforms continuously undergo rapid mutations in
     54 service of their need to maximize profit and expansion while minimizing the
     55 kinds of societal and individual harm that can plausibly cause them regulatory
     56 trouble or user disengagement. (The set of things that can cause trouble is
     57 also always shifting, as political and cultural spheres influence and are
     58 influenced by the platforms.) But platform mutations emerge only within a
     59 narrow range of possibilities delineated by the set of decisions considered
     60 valid in, roughly speaking, [11]Milton Friedman's model of corporate purpose.
     61 
     62 Within this circumscribed mutation zone, certain goals are able to be named and
     63 earnestly pursued ("stop spam" or "eliminate the distribution of CSAM"), even
     64 if they're never achieved. Other goals (anything to do with hate speech,
     65 incitement to violence, or misinformation, for example) can be named and
     66 pursued, but only in ways that don't hinder the workings of the
     67 profit-extraction machinery—which mostly means that they come in on the margins
     68 and after the fact, as in "[12]after the fact of a genocide that Facebook had
     69 years of explicit advance warnings about." Working on the margins and after the
     70 fact still matters—less damage is better than more damage—but it means "trust
     71 and safety" is kept well clear of the core.
     72 
     73 Again, this is all simple and obvious. A tractor structurally can't spare a
     74 thought for the lives of the fieldmice; shouting at the tractor when it
     75 destroys their nests is a category error. Business does business. The
     76 production line doesn't stop just because a few people lose fingers or lives.
     77 And what is a modern corporation but a legal spell for turning reasoning beings
     78 into temporarily vacant machines? We know this, which is why we have OSHA and
     79 the FAA and the FTC, for now.
     80 
     81 It's no surprise that when prodded by entities with cultural or regulatory
     82 power, platforms build more semi-effective AI classifiers, hire more underpaid
     83 contract moderators, and temporarily stiffen their unevenly enforced community
     84 rules, but then immediately slump back toward their natural form, which appears
     85 to be a cartoonishly overgrown early-2000s web forum crammed with soft targets
     86 and overrun by trolling, spam, and worse.
     87 
     88 It’s possible to make the argument that sufficiently strong leadership could
     89 make even a tech corporation appear to be capable of holding an ethical line,
     90 and maybe even capable of accepting slightly smaller profits in service of
     91 socially beneficial goals—and that, conversely, the awful people in charge are
     92 the main source of the problems. It’s not a very good argument, though, even
     93 when I make it myself.
     94 
     95 Yes, X is currently controlled by a bizarrely gibbering billionaire with
     96 obvious symptoms of late-stage Mad King disease. Yes, Facebook and
     97 Instagram—which control vastly more territory than X—are controlled by a
     98 feckless, Tulip-craze-mainlining billionaire with a long history of grudgingly
     99 up-regulating governance efforts when under public or governmental pressure and
    100 then immediately axeing them when the spotlight moves on. But would these
    101 platforms inflict less damage if they were led by people who valued the well
    102 being of others? Probably yes, to a degree. Twitter/X has offered a lurid
    103 natural experiment, and the changes in X after it moved from Jack Dorsey’s
    104 spacey techno-libertarian leadership to Elon Musk’s desperately needy
    105 quasi-fascist circus act have been obviously bad. A version of Meta founded and
    106 led by someone with a reasonably sharp ethical grounding clearly wouldn’t look
    107 much like the real Meta at all.
    108 
    109 On the other hand, TikTok’s social function is reasonably close to Meta and
    110 X’s, and the fact that [13]its CEO, Shou Zi Chew, seems like a relatively
    111 normal person, doesn’t seem to have correlated with dramatically better
    112 performance in eliminating [14]Nazi organizing, [15]genocidal and
    113 violence-inciting content, [16]CSAM distribution ([17]archive link), or the
    114 kind of [18]semi-pro disinformation that makes it harder for people
    115 experiencing natural disasters to understand what’s happening.
    116 
    117 Crucially, more reasonable CEO behavior doesn’t seem to prevent the lower-level
    118 and potentially even more destructive social effects of platforms that [19]
    119 Henry Farrell persuasively explains from a social theory perspective, or that
    120 Renée DiResta memorably calls a “Cambrian explosion of subjective, bespoke
    121 realities” in [20]Invisible Rulers.* (I'll do a separate post collecting
    122 thoughts on this angle, because it's too important to breeze by.)
    123 
    124 The realities of our moment also work against arguments for the potential of
    125 heroic leadership: even apparently level-headed tech executives now appear to
    126 understand that the next Trump Administration intends to rule unreasonably and
    127 vengefully, and that failure to [21]perform obeisance and [22]make tribute may
    128 result in federal interference that could plausibly unmake their companies.
    129 
    130 Those are not risks any global corporation can take, but our oddball lineup of
    131 big platform companies is in a special bind. No matter how desperately they
    132 want to be seen as neutral utilities, they have functioned, for good and ill,
    133 like social and political wrecking balls—and real or feigned misapprehensions
    134 about algorithms and censorship notwithstanding, real-world governments
    135 understand this. The second coming of Trump makes the situation especially
    136 stark, but the underlying dynamics are neither new nor temporary.
    137 
    138 Given that every large platform posing as a public square has put itself into
    139 the genuinely untenable situation of acting as a global corporate arbiter of
    140 politically hot speech, they will all always be in the gunsights of the world’s
    141 least reasonable governments. This was bad enough for the platforms when the
    142 least reasonable governments were Putin’s or Erdoğan’s or Modi’s—a truly
    143 unreasonable government in control of their home jurisdiction is an existential
    144 threat.
    145 
    146 And again, in reality, the corporations are configured to try to address the
    147 least political kinds of abuse—CSAM, spam, scams, and a few other forms of
    148 inauthentic behavior—and very little else. As a result, they can’t govern more
    149 subtle or politicized speech for much longer than I can roll a quarter down a
    150 piece of string.
    151 
    152 So what would it take for a corporation to become capable of good governance of
    153 things like political speech, incitement to violence and genocide, hate speech,
    154 most forms of inauthentic behavior, and platform manipulation? Two things, at
    155 least:
    156 
    157   • The ability and willingness to take and hold ethical stances that will be
    158     sharply unpopular with large swathes of the people mostly likely to
    159     effectively target them with legislation and abuses of power, and
    160   • the ability and willingness to devote something approaching the majority of
    161     their company’s time, money, and attention to building and running [23]
    162     devolved or [24]federated systems for doing high-performance high-context
    163     local governance according to those unpopular ethical stances.
    164 
    165 Can you bring yourself to imagine—concretely and in detail—these conditions
    166 occurring in the leadership of a global corporation?
    167 
    168 And again, achieving a mode of governance that can appropriately handle those
    169 most obvious elements—the hate speech, the network abuse, the inauthentic
    170 behavior, all of it—is necessary but not sufficient for reaching something like
    171 a healthy equilibrium. The elements of big social platforms that make them
    172 attractive and fun and profitable are the same elements that, as currently
    173 implemented, turn low-level human behavior patterns around status, belief,
    174 conformity, and predation into a high-speed mass-scale mess of fractured
    175 publics and realities.
    176 
    177 Two points of clarification: First, I’m not saying “Can’t fix people problems
    178 with technology,” which is exactly as true and useful as “Guns don’t kill
    179 people, people kill people.” (I used the former in what I thought was a very
    180 obviously sarcastic way, but apparently the intent was insufficiently clear.)
    181 If a technological system makes human problems worse, you have to fix the
    182 system or break it and build a better one.
    183 
    184 Second, none of what I’m trying to get at here is about the intent of people
    185 who work on big platforms. Corporate platform trust and safety staff routinely
    186 work themselves to the brink of individual illness or collapse to handle what
    187 they’re permitted and resourced to handle—which is itself a tiny fraction of
    188 what would be necessary to handle to make platforms good. Corporate platform
    189 governance by technology companies whose success requires growth and
    190 attention-extraction, though, is a bankrupt idea.
    191 
    192 If we briefly isolate the reality of our technological present, it’s hard to
    193 find it anything but absurd to expect a corporation to govern global or even
    194 local speech for any humanist value of “well.” And no one chose it, exactly, it
    195 just happened when the fantasies of the internet as an Apollonian zone of
    196 libertarian splendor met the reality of globally connected primate brains under
    197 late capitalism. I explicitly blame the connected-computer dream of
    198 technologically mediated liberation as cartoonishly exemplified in [25]JP
    199 Barlow’s Declaration, which centered on keeping the bad old world of human
    200 governance, which it equated with censorship, out of the internet:
    201 
    202     You claim there are problems among us that you need to solve. You use this
    203     claim as an excuse to invade our precincts. Many of these problems don't
    204     exist. Where there are real conflicts, where there are wrongs, we will
    205     identify them and address them by our means. We are forming our own Social
    206     Contract. This governance will arise according to the conditions of our
    207     world, not yours. Our world is different.
    208 
    209 The governance arose, all right, once the money got real.
    210 
    211 The computer dream’s rapidly evaporating and over-salinated shallows are still
    212 keeping the tech industry’s dumbest boats afloat, but the platforms have been
    213 scraping bottom for years while their owners slap on layers and layers of
    214 patches and bilge-pumps and bucket brigades manned by people from former
    215 colonies. The problem isn't (just) turning fact-checking on or off or
    216 deactivating a swarm of halfassed AI classifers or ceasing to pretend to act on
    217 most reports of misconduct, it's bad shape.
    218 
    219 All of which is to say that yes, Zuckerberg is a terrible chump and Musk is a
    220 grotesque quasi-Rasputin, and that does matter, but the boards they stand on
    221 have been rotten the whole time. Centralized corporate governance of global
    222 mega-platforms was always a goofy idea, and we should have given up on it years
    223 ago.
    224 
    225 This is where I get into awkward situations with lovely people, including
    226 several I count as friends, because they’re determined "not to let platforms
    227 off the hook.” I feel this, deeply, along with things like send the Sacklers to
    228 the guillotine. But keeping the fucked-up mutant fish on the hook will not
    229 magically transform it into an entity capable of governing.
    230 
    231    232 
    233 Earlier this week at Platformer, [26]Casey Newton reported some insider views
    234 on what Meta’s most recent roll-back of content moderation and fact-checking
    235 means. The post is worth reading, and after the Myanmar research I did in 2023,
    236 and for what it’s worth, I don’t think Casey’s sources overstate the dangers
    237 inherent in what Meta’s doing: more real human beings are going to suffer and
    238 lose children and be killed because of this. But I want to look at something in
    239 the cursory background section of the newsletter, about the work that Meta put
    240 in after 2016, when Facebook got criticized for hosting election-interference
    241 ops in the US:
    242 
    243     Chastened by the criticism, Meta set out to shore up its defenses. It hired
    244     40,000 content moderators around the world, invested heavily in building
    245     new technology to analyze content for potential harms and flag it for
    246     review, and became the world’s leading funder of third-party fact-checking
    247     organizations. It spent $280 million to create an independent Oversight
    248     Board to adjudicate the most difficult questions about online speech. It
    249     disrupted dozens of networks of state-sponsored trolls who sought to use
    250     Facebook, Instagram, and WhatsApp to spread propaganda and attack
    251     dissenters.
    252 
    253 [27]According to Financial Times reporting ([28]archive link), Meta currently
    254 employs or has contracted with about 40,000 people to work on “safety and
    255 security,” of which just 15,000 are content moderators, for a user base of
    256 roughly four billion users, which works out to more than a quarter of a million
    257 users per moderator. This chimes with [29]New York Times reporting ([30]archive
    258 link) suggesting that in 2021, Accenture was billing Facebook for about 5,800
    259 full-time contract moderators. (For what it’s worth, in 2017, Meta promised to
    260 add all of [31]3,000 trust and safety staff.) Nor are Meta’s moderation
    261 resources allocated evenly: About 90% of Facebook users are outside the US and
    262 Canada; that overwhelming majority gets [32]approximately 13% of the company’s
    263 moderation time ([33]archive link).
    264 
    265 And while we’re here, in 2020—the year Oversight Board started hiring—Meta
    266 cleared about $91 billion in profit. The Oversight Board trust got $280 million
    267 from Meta, or just over 0.3% of the company’s annual profits. The Oversight
    268 Board itself, though inclined to deliver thoughtful if glacially slow
    269 recommendations, appears to have accomplished [34]remarkably little.
    270 
    271 Again: The work tens of thousands of people around the world put in to try to
    272 make platforms less terrible is real and essential work, and it’s often done at
    273 a terrible cost. It’s also the barest gesture at serious governance, and much
    274 of it is pure Potemkin Village.
    275 
    276 That’s only a couple of things pulled from one paragraph that happened to hit
    277 my inbox while I was drafting this post, but I did [35]a whole lot of that kind
    278 of close reading in 2023, and came out believing that platform intensifications
    279 of governance in response to periodic governmental pressure are best understood
    280 as a little bit of real (though deeply inadequate) change and a whole lot of 
    281 [36]flopping. Then, when the pressure comes off, the platforms re-orient like
    282 compass needles tossed into in an MRI machine.
    283 
    284    285 
    286 [37]Barlow’s Declaration—which is excruciating and which I’ve been making
    287 myself reread annually for years as penance for participating in tech
    288 culture—ends like this:
    289 
    290     We will create a civilization of the Mind in Cyberspace. May it be more
    291     humane and fair than the world your governments have made before.
    292 
    293 What we got instead was a handful of global-scale company towns that continue
    294 to prove their comprehensive unfitness to govern and their absolute
    295 vulnerability to the offline governments the free internet was meant to work
    296 around.
    297 
    298 So sure: [38]Protocols over platforms. Then we have to actually do the
    299 inelegant, un-heroic, expensive work of rebuilding the essential structures of
    300 human civilization on top of the protocols, because it turns out we just have
    301 the one world, online or off, no way out.
    302 
    303 Thank you
    304 
    305 This post, like the others on this site, exists [39]because people have signed
    306 on to support the work. If you find it useful, and your situation allows for it
    307 with ease, please consider signing up! Enormous thanks to those of you who
    308 have. And a note to members: I've wrestled down the Ghost commenting problem
    309 and the first real discussion post for project members goes up tomorrow, so if
    310 you've signed up for a paid membership, look for that in your inbox soon.
    311 
    312 Notes
    313 
    314 * Renée’s book is very good and I recommend it for its lucid explanations and
    315 commitment to drawing on previous eras of mass communication without doing too
    316 deeply into either theoretical or historical rabbitholes (which I love, but
    317 which don’t make for popular reading). I don’t 100% agree with her conclusions,
    318 but they’re clearly stated and cleanly argued, which allows for productive
    319 disagreement—and I value that more than full alignment.
    320 
    321 A common response to the things I've been posting is "Okay, but what will work,
    322 then?" I think there are hints at answers in the very chunky [40]fediverse
    323 governance research I worked on last year, in online and offline cooperatives,
    324 in [41]Rudy Fraser's Blacksky, and in the kinds of projects Nathan Schneider
    325 assesses in [42]Governable Spaces. I'll continue to explore what I think might
    326 be good shapes for governance here in ways that—I hope—will be more pragmatic
    327 than quixotic.
    328 
    329 The featured diagram for this post is International Marine Engineering's 1912
    330 depiction of the profile and deck of the Titanic ([43]v. 17, p. 199).
    331 
    332 [44] [45] [46] The link has been copied!
    333 [47][https://www.wrecka.g]
    334 [48] How we'll do discussions here
    335 Newer post
    336 
    337 [49]How we'll do discussions here
    338 
    339 Older post
    340 
    341 [50]What people in the global majority need from networks
    342 
    343 [51] What people in the global majority need from networks
    344 
    345 Subscribe to new posts.
    346 
    347 [52][                    ] Subscribe
    348 Processing your application Great! Check your inbox and confirm your
    349 subscription There was an error sending the email
    350 choose your type:
    351 
    352   • [54]( ) fancy (default)
    353   • [55]( ) simpler
    354 
    355 [56]wreckage/salvage
    356 
    357 Making and mending networks for humans.
    358 
    359 wreckage/salvage © Erin Kissane 2025  
    360 You’ve successfully subscribed to wreckage/salvage
    361 Welcome back! You’ve successfully signed in.
    362 Great! You’ve successfully signed up.
    363 Success! Your email is updated.
    364 Your link has expired
    365 Success! Check your email for magic link to sign-in.
    366 
    367 References:
    368 
    369 [1] https://www.wrecka.ge/
    370 [3] https://www.wrecka.ge/about/
    371 [4] https://www.wrecka.ge/tag/posts/
    372 [5] https://www.wrecka.ge/#/portal
    373 [6] https://www.wrecka.ge/signin/
    374 [7] https://www.wrecka.ge/author/erin/
    375 [8] https://senseis.xmp.net/?LifeAndDeath
    376 [9] https://www.gutenberg.org/files/46/46-h/46-h.htm
    377 [10] https://www.theengineroom.org/library/new-report-exploring-a-transition-to-alternative-social-media-platforms-for-social-justice-organizations-in-the-majority-world/
    378 [11] https://www.nytimes.com/1970/09/13/archives/a-friedman-doctrine-the-social-responsibility-of-business-is-to.html
    379 [12] https://erinkissane.com/meta-in-myanmar-part-i-the-setup
    380 [13] https://www.wired.com/story/shou-zi-chew-tik-tok-big-interview/
    381 [14] https://www.isdglobal.org/digital_dispatches/naztok-an-organized-neo-nazi-tiktok-network-is-getting-millions-of-views/
    382 [15] https://www.isdglobal.org/wp-content/uploads/2021/08/HateScape_v5.pdf
    383 [16] https://www.ft.com/content/d5549d48-8f02-464c-9c7d-17404a5b6d02
    384 [17] https://archive.ph/tvn8K
    385 [18] https://newrepublic.com/article/186928/misinformation-new-normal-disaster-response
    386 [19] https://www.programmablemutter.com/p/were-getting-the-social-media-crisis
    387 [20] https://www.hachettebookgroup.com/titles/renee-diresta/invisible-rulers/9781541703377/
    388 [21] https://www.newsweek.com/tech-ceos-donations-donald-trump-joe-biden-inaugurations-compared-2010457
    389 [22] https://qz.com/google-youtube-trump-inauguration-meta-amazon-apple-1851736124
    390 [23] https://en.wikipedia.org/wiki/Devolution
    391 [24] https://en.wikipedia.org/wiki/Federalism
    392 [25] https://www.eff.org/cyberspace-independence
    393 [26] https://www.platformer.news/meta-fact-checking-free-speech-surrender/?ref=platformer-newsletter
    394 [27] https://www.ft.com/content/afeb56f2-9ba5-4103-890d-91291aea4caa
    395 [28] https://archive.ph/UxgFy
    396 [29] https://www.nytimes.com/2021/08/31/technology/facebook-accenture-content-moderation.html
    397 [30] https://archive.ph/SEZpr
    398 [31] https://www.nytimes.com/2017/05/03/technology/facebook-moderators-q1-earnings.html
    399 [32] https://www.wsj.com/articles/facebook-drug-cartels-human-traffickers-response-is-weak-documents-11631812953
    400 [33] https://archive.ph/https://www.wsj.com/articles/facebook-drug-cartels-human-traffickers-response-is-weak-documents-11631812953
    401 [34] https://www.lawfaremedia.org/article/meta-s-oversight-board-and-the-need-for-a-new-theory-of-online-speech
    402 [35] https://erinkissane.com/meta-in-myanmar-part-iii-the-inside-view
    403 [36] https://www.youtube.com/watch?v=Ml8kf3UIpN0
    404 [37] https://www.eff.org/cyberspace-independence
    405 [38] https://knightcolumbia.org/content/protocols-not-platforms-a-technological-approach-to-free-speech
    406 [39] https://www.wrecka.ge/#/portal
    407 [40] https://fediverse-governance.github.io/
    408 [41] https://www.wired.com/story/blacksky-is-nothing-like-black-twitter/
    409 [42] https://www.ucpress.edu/books/governable-spaces/epub-pdf
    410 [43] https://archive.org/details/internationalma171912newy/page/198/mode/2up
    411 [44] https://tootpick.org/#text=%22Bad%20shape%22%20https://www.wrecka.ge/bad-shape/
    412 [45] https://bsky.app/intent/compose?text=%22Bad%20shape%22+https://www.wrecka.ge/bad-shape/
    413 [46] javascript:
    414 [48] https://www.wrecka.ge/how-we-do-it-here/
    415 [49] https://www.wrecka.ge/how-we-do-it-here/
    416 [50] https://www.wrecka.ge/what-people-in-the-global-majority-need-from-networks/
    417 [51] https://www.wrecka.ge/what-people-in-the-global-majority-need-from-networks/
    418 [56] https://www.wrecka.ge/