paulgraham-com-9dskkh.txt (54352B)
1 bg image(https://s.turbifycdn.com/aah/paulgraham/bel-6.gif) 2 [bel-7] * [2][bel-8] 3 4 The Brand Age 5 6 March 2026 7 8 In the early 1970s disaster struck the Swiss watch 9 industry. Now people call it the quartz crisis, but in 10 fact it was a compound of three separate disasters 11 that all happened at about the same time. 12 13 The first was competition from Japan. The Swiss had 14 been watching the Japanese in the rear view mirror all 15 through the 1960s, and they'd been improving at an 16 alarming rate. But even so the Swiss were surprised in 17 1968 when the Japanese swept all the top spots for 18 mechanical watches at the Geneva Observatory trials. 19 20 The Swiss knew what was coming. For years the Japanese 21 had been able to make cheaper watches. Now they could 22 make better ones too. 23 24 To make matters worse, Swiss watches were about to 25 become much more expensive. The Bretton Woods 26 agreement, which since 1945 had fixed the exchange 27 rates of most of the world's currencies, had set the 28 Swiss Franc at an artificially low rate of .228 USD. 29 When Bretton Woods collapsed in 1973, the Franc shot 30 upward. By 1978 it reached .625 USD, meaning Swiss 31 watches were now 2.7 times as expensive for Americans 32 to buy. [[3]1] 33 34 The combined effect of foreign competition and the 35 loss of their protective exchange rate would have 36 decimated the Swiss watch industry even if it hadn't 37 been for quartz movements. But quartz movements were 38 the final blow. Now the whole game they'd been trying 39 to win at became irrelevant. Something that had been 40 expensive — knowing the exact time — was now a 41 commodity. 42 43 Between the early 1970s and the early 1980s, unit 44 sales of Swiss watches fell by almost two thirds. Most 45 Swiss watchmakers became insolvent or close to it and 46 were sold. But not all of them. A handful survived as 47 independent companies. And the way they did it was by 48 transforming themselves from precision instrument 49 makers into luxury brands. 50 51 In the process the nature of the mechanical watch was 52 also transformed. The most expensive watches have 53 always cost a lot, but why they cost a lot and what 54 buyers got in return have changed completely. In 1960 55 expensive watches cost a lot because they cost a lot 56 to manufacture, and what the buyer got in return was 57 the most accurate timekeeping device, for its size, 58 that could be made. Now they cost a lot because brands 59 spend a lot on advertising and use tricks to limit 60 supply, and what the buyer gets in return is an 61 expensive status symbol. 62 63 That turns out to be a profitable business though. The 64 Swiss watch industry probably makes more now from 65 selling brand than they would have if they were still 66 selling engineering. And indeed, when you look at the 67 graph of Swiss watch sales by revenue, it tells a 68 different story than the graph of unit sales. Instead 69 of falling off a cliff, the revenue numbers merely 70 flatten out for a while, and then take off like a 71 rocket in the late 1980s as the surviving watchmakers 72 come to terms with their new destiny. 73 74 It took the watchmakers about 20 years to figure out 75 the new rules of the game. And it's interesting to 76 watch them do it, because the completeness of their 77 transformation makes it the perfect case study in one 78 of the most powerful forces of our era: brand. 79 80 Brand is what's left when the substantive differences 81 between products disappear. But making the substantive 82 differences between products disappear is what 83 technology naturally tends to do. So what happened to 84 the Swiss watch industry is not merely an interesting 85 outlier. It's very much a story of our times. 86 87 88 89 90 91 Jaeger-LeCoultre's web site says that one of their 92 current collections "takes its inspiration from the 93 classic designs of the golden age of watchmaking." In 94 saying this they're implicitly saying something that 95 present-day watchmakers all know but rarely come so 96 close to saying outright: whatever age we're in now, 97 it's not the golden age. 98 99 The golden age was from 1945 to 1970 — from the point 100 where the watch industry emerged from the chaos of war 101 with the Swiss on top till the triple cataclysm that 102 struck it starting in the late 60s. There were two 103 things watchmakers sought above all in the golden age: 104 thinness and accuracy. And indeed this was arguably 105 the essential tradeoff in watchmaking. A watch is 106 something you carry with you to tell you the time. So 107 there are two fundamental ways to improve it: to make 108 it easier to carry with you and to make it better at 109 telling the time. 110 111 Obviously accuracy is valuable, but in the golden age 112 thinness was if anything more valuable. Even in the 113 days of pocket watches the best watchmakers tried to 114 make their watches as thin as they could. Cheap, thick 115 pocket watches were derided as "turnips." But thinness 116 took on a new urgency when men's watches moved onto 117 their wrists during World War I. And since thinness 118 was more difficult to achieve than accuracy, it was 119 this quality that tended to distinguish the more 120 expensive watches of the golden age. 121 122 There is one other thing watchmakers have pursued in 123 some eras: telling more than the time in the usual 124 way. Telling you the phase of the moon, for example, 125 or telling the time with sound. In the industry the 126 term for these things is "complications." They were 127 popular in the nineteenth century and they're popular 128 again now, but except for one pragmatic complication 129 (showing the date), they were a sideshow in the golden 130 age. In the golden age, as always in golden ages, the 131 top watchmakers focused on the essential tradeoff. 132 And, as always in golden ages, they did it 133 beautifully. The best watches of the golden age have a 134 [4]quiet perfection that has never been equalled 135 since. And for reasons I'm about to explain, probably 136 never will be. 137 138 The three most prestigious brands of the golden age 139 were the so-called "holy trinity" of Patek Philippe, 140 Vacheron Constantin, and Audemars Piguet. Their 141 prestige was mostly deserved; they had earned it by 142 the exceptional quality of their work. By the 1960s 143 they stood on two legs, prestige and performance. And 144 what they learned in the next two decades was that 145 they had to put all their weight on the first leg, 146 because they could no longer win at either of the two 147 things watchmakers had historically striven to 148 achieve. Quartz movements were not only more accurate 149 than any mechanical movement, but thinner too. 150 151 The holy trinity at least had another leg to stand on. 152 Most of the other well-known Swiss watchmakers sold 153 only performance. None of those companies survived 154 intact. 155 156 Omega showed what not to do. Omega were the nerds of 157 Swiss watchmakers. They made wonderfully accurate 158 watches, but they would have been ambivalent, at best, 159 about the idea of being a luxury brand. When the 160 Japanese got as good as the Swiss at making accurate 161 movements, Omega responded in the Omega way: make even 162 more accurate movements. They introduced a new 163 movement in 1968 that ran at a 45% higher frequency. 164 In theory this should have made it more accurate, but 165 the new movement was so fragile that it destroyed 166 their reputation for reliability. They even tried to 167 make a better quartz movement, but there was nothing 168 down that road but a race to the bottom. By 1981 they 169 were insolvent and were taken over by their creditors. 170 171 Patek Philippe took the opposite approach. While Omega 172 was redesigning their movements, Patek was redesigning 173 their cases. Or more precisely, designing their cases, 174 because until then they hadn't. 175 176 This is probably the point to mention what a strange 177 beast the Swiss watch industry was in those days. It 178 was a kind of capitalism that's hard to imagine today, 179 and even then could only have been made to work in a 180 country like Switzerland — a network of small, 181 specialized companies locked into place by regulation. 182 The companies that we for convenience have been 183 calling watchmakers were merely the consumer-facing 184 edge of this network. The holy trinity didn't design 185 their own cases, or even their own movements most of 186 the time. 187 188 In 1968 (that year again) Patek Philippe launched a 189 new watch that shifted the center of gravity of case 190 design. This time they'd taken their own designs to 191 the casemakers and said "this is what you're going to 192 make for us." The result was a striking new model 193 called the [5]Golden Ellipse. Somewhat confusingly, 194 because it wasn't elliptical. The new case was more of 195 what UI designers would call a round rect: a rectangle 196 with rounded corners. And this new family of watches 197 was quite successful. But it was more than that: it 198 was the pattern for the future. [[6]2] 199 200 How could merely designing a distinctive case be so 201 important? Because it turned the entire watch into an 202 expression of brand. 203 204 The trouble with the best watches of the golden age, 205 from the point of view of someone who wanted to 206 impress people with the brand of watch he was wearing, 207 was that no one could tell what brand of watch you 208 were wearing. Until you got within a few inches of 209 them, the watches of all the top makers looked the 210 same. That's the thing about minimalism: there tends 211 to be just one answer. Plus the watches of the golden 212 age were small by present standards. Watchmakers had 213 spent centuries working to make them smaller, and by 214 1960 they'd gotten very good at it. So the only thing 215 distinguishing one top brand from another was the name 216 printed on the dial, and dials were so small that 217 these names were tiny. The manufacturers' names on the 218 holy trinity's golden age watches are between half and 219 three quarters of a millimeter high. So by taking over 220 the case, Patek expanded the size of the brand from 8 221 square millimeters to 800. 222 223 Why did they suddenly decide to make their brand 224 shout, after a century of whispering? Because they 225 knew they weren't going to beat the Japanese on 226 performance. From now on they'd have to depend more on 227 brand. 228 229 There's a cost to doing this, which we can see even in 230 this early example of case-as-brand. Golden Ellipses 231 are not bad looking. They must have looked even cooler 232 in the 1970s, when designers were turning everything 233 into round rects. But the Golden Ellipse was not an 234 evolutionary step forward in case design. Watches 235 didn't all become round rects. Watchmakers had already 236 discovered the optimal shape for the case of something 237 that describes a circle as it rotates. 238 239 They had also discovered the optimal shape for the 240 crown, the knob on the side of a watch that you turn 241 to wind it. But to emphasize the distinctive profile 242 of the Ellipse, Patek made the crown too small, with 243 the result that they're distractingly hard to wind. [ 244 [7]3] 245 246 So even in this early example we see an important 247 point about the relationship between brand and design. 248 Branding isn't merely orthogonal to good design, but 249 opposed to it. Branding by definition has to be 250 distinctive. But good design, like math or science, 251 seeks the right answer, and right answers tend to 252 converge. 253 254 Branding is centrifugal; design is centripetal. 255 256 There is some wiggle room here of course. Design 257 doesn't have as sharply defined right answers as math, 258 especially design meant for a human audience. So it's 259 not necessarily bad design to do something distinctive 260 if you have honest motives. But you can't evade the 261 fundamental conflict between branding and design, any 262 more than you can evade gravity. 263 264 Indeed, the conflict between branding and design is so 265 fundamental that it extends far beyond things we call 266 design. We see it even in religion. If you want the 267 adherents of a religion to have customs that set them 268 apart from everyone else, you can't make them do 269 things that are convenient or reasonable, or other 270 people would do them too. If you want to set your 271 adherents apart, you have to make them do things that 272 are inconvenient and unreasonable. 273 274 It's the same if you want to set your designs apart. 275 If you choose good options, other people will choose 276 them too. 277 278 There are only two ways to combine branding and good 279 design. You can do it when the space of possibilities 280 is enormously large, as it is in painting for example. 281 Leonardo could paint as well as he possibly could and 282 yet also paint in a style that was distinctively his. 283 If there had been a million painters as good as 284 Bellini and Leonardo this would have been harder to 285 do, but since there were more like ten they didn't 286 bump up against one another much. [[8]4] 287 288 The other situation when branding and good design can 289 be combined is when the space of possibilities is 290 comparatively unexplored. If you're the first to 291 arrive in some new territory, you can both find the 292 right answer and claim it as uniquely yours. At least 293 at first; if you've really found the right answer, 294 everyone else's designs will inevitably converge on 295 yours, and your brand advantage will erode over time. 296 297 Since the space of watch design is neither unexplored 298 nor enormously large, branding can only be achieved at 299 the expense of good design. And in fact if you wanted 300 one sentence to describe the current age of 301 watchmaking, that one would do pretty well. 302 303 Patek Philippe didn't know for sure that making 304 visibly branded watches would work. It was not even 305 their only strategy, at the time. They were finding 306 their way. But it was the strategy that did work, at 307 least as measured by revenues. 308 309 For it to work the customers had to meet them halfway. 310 Patek knew that not all their customers were buying 311 their watches for the performance they delivered — for 312 their accuracy and thinness. They knew that at least 313 some customers were buying them because they were 314 expensive. But it was unclear how many, or how far 315 they could be pushed. 316 317 To encourage them, Patek did something that none of 318 the holy trinity had done much of before: brand 319 advertising. And what they talked about was how 320 expensive their watches were. A 1968 Patek ad 321 explained "why you are well advised to invest perhaps 322 half a month's income" in an Ellipse. "Like every 323 Patek Philippe," the ad continued, "this thin model is 324 entirely finished by hand. Since a Patek Philippe is 325 the costliest watch to make, production is severely 326 limited: only 43 watches are signed out each day for 327 delivery to prominent jewelers throughout the world." 328 [[9]5] 329 330 You can tell this is an early ad because they still 331 mention thinness. But there is no mention of accuracy. 332 Presumably Patek felt that battle was already lost. 333 334 The next move was made by Audemars Piguet, who in 1970 335 commissioned the renowned designer Gérald Genta to 336 design their own iconic watch, this one, daringly, in 337 steel. The result, launched in 1972, was the [10]Royal 338 Oak. And Audemars Piguet's ads (for they too now 339 started doing brand advertising) emphasized its high 340 cost even more dramatically. "Introducing steel at the 341 price of gold," one began. "You're looking at the 342 costliest stainless steel watch in the world — the 343 Audemars Piguet 'Royal Oak'. What makes it even more 344 precious than gold is the time that went into building 345 it, by a vanishing breed of master watchmakers." At 346 the bottom of the ad they turn the traditional formula 347 on its head and describe their watches as being 348 "priced from $35,000 and down." 349 350 The Royal Oak was also a step forward in surface area 351 devoted to brand. The Golden Ellipse had turned the 352 watch face into an expression of brand, but it used 353 ordinary straps and bracelets. In the Royal Oak, the 354 watch face was integrated with a metal bracelet that 355 continued its design all the way around the wrist. 356 When it said "You're looking at the costliest 357 stainless steel watch in the world," it said it with 358 every square millimeter of surface area. 359 360 Would customers buy this new approach? The initial 361 results were moderately encouraging. The holy 362 trinity's sales didn't take off, but they didn't go 363 down to zero either. There were at least some people 364 out there responding to the new message. Perhaps if 365 they kept at it the number would grow. 366 367 So they did. Encouraged by the success of the Royal 368 Oak, Patek Philippe commissioned Gérald Genta in 1974 369 to design a similar watch for them. The design of the 370 Royal Oak had been inspired by a ship's porthole, so 371 the design of this new watch would be inspired by... a 372 ship's porthole. It was called the [11]Nautilus, and 373 it launched at the Basel Watch Fair in 1976. 374 375 In the Nautilus we really see the incompatibility of 376 branding and design. It was huge. The most expensive 377 men's watches at the peak of the golden age were 378 typically 32 or 33 millimeters in diameter. The 379 Nautilus was 42 millimeters. And as well as being huge 380 it had gratuitous knobs on either side of the face, 381 like a pair of ears. But you could recognize one from 382 across the room. 383 384 Of all the watches Patek makes now, the Nautilus is 385 the most sought after. It's perfectly aligned with 386 what present-day buyers want — basically, the loudest 387 possible expression of brand. But in 1976 it was ahead 388 of its time. In 1976 it was still a little too much. 389 390 The watch that finally turned Patek's fortunes around 391 was another iconic design, the hobnail calatrava. The 392 hobnail calatravas were so called because they were 393 decorated with tiny pyramid-shaped spikes. That was 394 enough to make them look distinctive. But except for 395 the hobnails they were basically golden age dress 396 watches. 397 398 The hobnail calatrava was apparently the brainchild of 399 René Bittel, the head of Patek Philippe's ad agency. 400 It was not a new design. Many watchmakers had 401 decorated their cases with hobnails over the years, 402 and there had been a Patek model with them since 1968. 403 But in 1984 Bittel told Patek president Philippe 404 Stern, in effect: make this your standard design, and 405 I'll create an ad campaign to identify it in people's 406 heads with your brand. [[12]6] 407 408 It worked spectacularly well. The resulting watch, the 409 [13]3919, is known as the "banker's watch" because it 410 became so popular among investment bankers in New York 411 in the 80s and 90s. Up to this point Patek had been 412 hedging their bets, making quartz watches as well, and 413 arguing defensively in their ads that quartz watches 414 in fancy cases were almost as laborious to make as 415 mechanical ones. But the ibankers bought the full 416 mechanical story. They didn't even need self-winding 417 mechanical watches; the 3919 was hand-wound. So be it. 418 Patek stopped talking about quartz movements. And 419 their sales, which had been flat since the early 70s, 420 were by 1987 on a clear upward trajectory that has 421 continued to this day. 422 423 It's hard to say for sure whether the critical 424 ingredient was Bittel's skill at advertising or a 425 receptive audience, but as someone who knew these 426 investment bankers, I'd lean toward the audience. 427 These were the people for whom the term "yuppy" was 428 coined. Living expensively was one of the things they 429 were best known for. If anyone was going to adopt a 430 new way to display wealth, it would be them. Whereas 431 if Bittel had sent the same message ten years earlier, 432 there might have been no one to hear it. 433 434 Whatever the cause, something happened in the second 435 half of the 1980s, because that's when all the numbers 436 finally start going up again. Up till about 1985 it 437 was still not clear what would happen with mechanical 438 watches. By 1990 it was. By 1990 the custom of using 439 expensive, highly-branded, conspicuously mechanical 440 watches as status symbols was firmly established. [ 441 [14]7] 442 443 Obsolete technologies don't usually get adopted as 444 ways to display wealth. Why did it happen with 445 mechanical watches? Because the wristwatch turns out 446 to be the perfect vehicle for it. Where better than 447 right on your wrist, where everyone can see it? And 448 more to the point, what better to do it with? You 449 could wear a diamond ring or a gold chain, but those 450 would have seemed socially dubious to investment 451 bankers. They might have been barbarians, but they 452 weren't mafia. Whereas nothing could be more legit 453 than a gold watch. The chairman of the company was 454 still wearing one his wife gave him 20 years ago, 455 before quartz watches were even a thing. If the 456 increasing pressure to display wealth was going to 457 emerge anywhere, this was the place. [[15]8] 458 459 For men, at least. Women never really went for the 460 idea of wearing mechanical watches. Most rich women 461 are happy wearing a Cartier tank with a quartz 462 movement. Why the difference? Partly for the same 463 reason that most buyers of steam engines are men. But 464 the main reason is that expensive mechanical watches 465 now serve as de facto jewelry for men, and women don't 466 need de facto jewelry because they can wear actual 467 jewelry. 468 469 It was critical, though, that mechanical watches were 470 accurate enough. A new 3919 would have been off by no 471 more than 5 seconds a day. That was nowhere near as 472 good as quartz. Even the cheapest mass market quartz 473 watches were accurate to half a second a day, and the 474 best ones were accurate to 3 seconds a year. But in 475 practice you didn't need that kind of accuracy. If 476 mechanical watches had only been accurate to a minute 477 a day they couldn't have made the leap from keeping 478 time to displaying wealth. It would have seemed too 479 manifestly unluxurious to have a watch that always had 480 the wrong time. But 5 seconds a day was close enough. 481 [[16]9] 482 483 This is an important point about the relationship 484 between brand and quality. Quality doesn't stop 485 mattering when a product switches to something people 486 buy for its brand. But the way it matters changes 487 shape. It becomes a threshold. It no longer has to be 488 so great that it sells the product; brand sells the 489 product; but it does have to be good enough to 490 maintain the brand's reputation. The brand must not 491 break character. 492 493 It was a lucky thing for the watchmakers that yuppies 494 arose just in time to save them. Or maybe not so 495 lucky. Because the evolution of the market that 496 yuppies represented has continued with a vengeance, 497 and watchmakers have perforce been dragged along with 498 it. If they don't make gigantic blingy watches for 499 buyers in Hong Kong and Dubai, someone else will. So 500 that is what they now find themselves doing. And what 501 began with a few comparatively subtle examples of the 502 conflict between branding and design is now an all out 503 [17]war on design. 504 505 The present era of mechanical watchmaking doesn't yet 506 have a name. But if we need one, it's obvious what it 507 should be: the brand age. The golden age ran from 1945 508 to 1970, followed by the quartz crisis from 1970 to 509 1985. Since 1985 we've been in the brand age. 510 511 512 513 514 515 This won't be the only brand age. Indeed, it's not 516 even the first; fine art has been in its own brand age 517 since the establishment of the Barr canon in the 518 1930s. And since we'll probably see more of this kind 519 of thing, it would be worth taking some time to look 520 at what a brand age is like. 521 522 How are things different now from the way they were in 523 the golden age? The best way to answer that might be 524 to imagine what someone from the golden age would 525 notice if we brought him here in a time machine. 526 527 The first thing he'd notice, if he walked through a 528 fancy shopping district, is that all the prominent 529 watchmakers of the golden age seem to be doing better 530 than ever. They're not only all still around, but most 531 now have their own boutiques instead of depending on 532 jewelers to sell their products as they used to back 533 in the day. 534 535 In fact this is an illusion. Only three watchmakers 536 survived the dark days of the 70s and 80s as 537 independent companies: Patek Philippe, Audemars 538 Piguet, and Rolex. All the rest are owned by six 539 holding companies, which reinflated them as it became 540 clear that mechanical watches would have a second life 541 as luxury accessories for men. Instead of separate 542 companies they're now more like the brands that got 543 rolled up into the big three American automakers: 544 they're ways for their parent companies to target 545 different segments of the market. So Longines, for 546 example, no longer competes with Omega, because the 547 company that owns them both has assigned it a lower 548 tier of the market. [[18]10] 549 550 There's a reason the Vacheron Constantin boutique 551 looks so much like the IWC and Jaeger-LeCoultre 552 boutiques, and for that matter the Montblanc and 553 Cartier boutiques. They're all owned by the same 554 company. It's similar with clothing brands, 555 incidentally. When you walk through a town's fanciest 556 shopping district, what seem to be the shops of lots 557 of different brands are actually owned by a handful of 558 conglomerates. That's one reason these districts seem 559 so sterile; like suburbs built by a single developer, 560 they have an unnatural lack of variety. 561 562 When our time traveler peered into the windows of 563 these shops, the first thing he'd notice was how large 564 all the watches were. This would surprise him, because 565 in the golden age, as indeed in all the preceding 566 centuries, big meant cheap. An expensive golden age 567 men's watch might have been 33 millimeters in diameter 568 and 8 millimeters thick. An expensive watch today will 569 be more like 42 millimeters in diameter and 10 570 millimeters thick — more than double the size. It 571 would astonish our visitor to look through the windows 572 of what were clearly very fancy shops and see what 573 seemed to be cheap watches. [[19]11] 574 575 We know how this happened. When watches switched from 576 telling time to telling brand, they grew in size to be 577 better at it. And not just in size, but in shape too. 578 That's another thing our time traveler would notice: 579 the surprising variety of strange case shapes and 580 awkward protrusions that have been produced as the 581 centrifugal tendency of branding played out. What, 582 he'd wonder, is going on with the huge guards on the 583 crowns of those Panerais? What do people do with these 584 watches that makes the crown need such protection? And 585 why would a crown guard have a message engraved on it 586 saying that it's a registered trademark? It's obvious 587 to us what's going on here, but imagine how confusing 588 it would be to someone from the golden age, when form 589 followed function. [[20]12] 590 591 As he puzzled over this strange assortment of bulky 592 watches, he'd notice a further pattern. He'd realize 593 that a surprisingly large number of them looked like a 594 specific brand of bulky watch he was already familiar 595 with. 596 597 I haven't talked about Rolex so far, because Rolex 598 didn't have to do much to adapt to the new era. They 599 already had one foot in the brand age during the 600 golden age. Early in their history they put a lot of 601 effort into making their watches better, but they 602 "stopped taking part in competitions in Geneva and 603 Neuchâtel at the end of the 1950s," and from about 604 1960 "largely abandoned research into mechanical 605 watchmaking." [[21]13] The reason was not that they'd 606 become lazy, but that they'd discovered they could 607 make sales grow faster by marketing their watches as 608 status symbols. So that became their focus during the 609 1960s, and by the time the quartz crisis hit ten years 610 later, their customers were self-selected to be people 611 who didn't care that much what was inside a watch, so 612 long as it was recognizably a Rolex. 613 614 And they were far ahead of other watchmakers in that 615 department. They already had in the 1940s what we saw 616 Patek Philippe and Audemars Piguet struggling to 617 create in the 1970s and 80s: a case that immediately 618 proclaimed the brand of the maker. The Rolex look 619 seems to have evolved organically, but once it did, 620 they realized how important it was. In fact they 621 pitched it as one of the features of their watches. A 622 1960s Rolex ad says "You can recognize its classic 623 shape, carved out of a block of solid gold, from the 624 other end of the conference table." 625 626 Indeed Rolex was ahead of its time in both dimensions: 627 their cases were not merely recognizable, but big too, 628 at least by golden age standards. That was not the 629 result of clever marketing, though. It was a byproduct 630 of the founder Hans Wilsdorf's obsession with building 631 waterproof watches. 632 633 As its name suggests, that was the raison d'etre of 634 the Rolex Oyster. Watches like the Oyster were 635 designed to be tough, like Jeeps. In the golden age 636 there were two poles of watch design. At one end were 637 tool watches, which were thick, tough, and usually 638 made of steel. At the other end were dress watches, 639 which were thin, elegant, and usually made of gold. 640 But Rolex blurred the line between them. When they 641 made thick, tough watches, they made them out of gold 642 as well as steel. The result was a sort of luxury 643 Jeep. And if that phrase didn't ring a bell in your 644 head, stop and think about it, because that is exactly 645 what everyone is driving now. That's what SUVs are, 646 luxury Jeeps. What happened to watches is the same 647 thing that happened to cars. And indeed if our time 648 traveler turned and saw a Porsche Cayenne pass by and 649 realized what it was — a huge, pseudo-offroad vehicle 650 meant to recall the Porsche 911 — he might have been 651 even more shocked than he was by the watches he'd been 652 looking at. [[22]14] 653 654 If the time traveller walked into a Patek Philippe 655 boutique and actually tried to buy a Nautilus, he'd 656 get the biggest shock of all. They wouldn't sell him 657 one. Because at Patek he'd encounter the most extreme 658 brand age phenomenon: artificial scarcity. You can't 659 just buy a Nautilus. You have to spend years proving 660 your loyalty first by buying your way through multiple 661 tiers of other models, and then spend years on a 662 waiting list. [[23]15] 663 664 Obviously this strategy sells more watches. But it 665 also supports retail prices by keeping watches off the 666 secondary market. A company using artificial scarcity 667 to drive sales can't allow too many of the scarce 668 models to leak into the secondary market, or they stop 669 being scarce. The ideal is the watch equivalent of 670 carbon sequestration: for the people who buy their 671 watches to keep them till they die. 672 673 To push the market toward this ideal, Patek squeezes 674 from both sides of the sale. They weed out flippers by 675 making the path to the scarce models so costly in both 676 time and money — so inconvenient and unreasonable — 677 that only a genuine fan would endure it. The lower 678 tier watches sell for below retail on the secondary 679 market, because Patek doesn't restrict their supply, 680 so a would-be flipper should have to spend years 681 making money-losing purchases before he could even get 682 something he could flip at a profit. Apparently some 683 people still manage to beat this system though, so 684 Patek's countermeasures don't end there. They keep a 685 vigilant eye on secondary sales to see who's selling 686 their watches. Auction listings usually include serial 687 numbers, so those are easy to trace, but if necessary 688 they'll rebuy their own watches on the secondary 689 market to get the serial number and trace the leak. 690 They buy hundreds a year. And when they catch someone 691 selling watches they don't want them to, they don't 692 just cut off that customer. If a retailer's customers 693 are responsible for too many such leaks, they'll cut 694 off the whole retailer. Which naturally makes 695 retailers eager to help them police buyers. 696 697 There will of course always be some leaks into the 698 secondary market. Even the most loyal customers die at 699 a certain rate. And in fact it's critical for Patek 700 that the secondary market continue to exist, because 701 it's one of the most valuable sources of information 702 they have about the most important question they face: 703 how fast to increase the supply of the top tier 704 watches. Their scarcity helps drive the purchases of 705 all the others, so those that do make it into the 706 secondary market should always sell for above retail. 707 And I'm sure Patek leaves a large margin for error 708 when increasing supply, because if secondary market 709 prices for these watches get close to retail prices, 710 you're getting close to a price collapse — which, 711 since people now buy these watches as investments, 712 would have the same disastrous cascading effect as the 713 bursting of an asset bubble. It wouldn't just be like 714 the bursting of an asset bubble. It would be the 715 bursting of an asset bubble. That's the business an 716 elite watchmaker is in now: carefully managing a 717 sustained asset bubble. [[24]16] 718 719 This is an instance of what I call the comb-over 720 effect: when a series of individually small changes 721 takes you from something that's a little bit off to 722 something that's freakishly wrong. I'm sure Patek 723 didn't cook up this whole scheme in one shot; I'm sure 724 it evolved gradually. But look at what a strange place 725 we've ended up in. Back in the golden age the way you 726 bought a Patek Philippe was to go to a jeweler and 727 give them money. Now Patek is policing buyers to 728 maintain an asset bubble. 729 730 The most striking thing to me about the brand age is 731 the sheer strangeness of it. The zombie watch brands 732 that appear to be independent and even have their own 733 retail stores, and yet are all owned by a few holding 734 companies. The giant, awkwardly shaped watches that 735 reverse 500 years of progress in making them smaller. 736 The business model that requires a company to rebuy 737 their own watches on the secondary market to catch 738 rogue customers. The very concept of rogue customers. 739 It's all so strange. And the reason it's strange is 740 that there's no function for form to follow. 741 742 Up to the end of the golden age, mechanical watches 743 were necessary. You needed them to know the time. And 744 that constraint gave both the watches and the 745 watchmaking industry a meaningful shape. There were 746 certainly some strange-looking watches made during the 747 golden age. They weren't all beautifully minimal. But 748 when golden age watchmakers made a strange-looking 749 watch, they knew they were doing it. In fact they give 750 the impression of having done it as a deliberate 751 exercise, to avoid getting into a rut. 752 753 That's not why brand age watches look strange. Brand 754 age watches look strange because they have no 755 practical function. Their function is to express 756 brand, and while that is certainly a constraint, it's 757 not the clean kind of constraint that generates good 758 things. The constraints imposed by brand ultimately 759 depend on some of the worst features of human 760 psychology. So when you have a world defined only by 761 brand, it's going to be a weird, bad world. 762 763 764 765 766 767 Well that was dark. Is there some edifying lesson we 768 can salvage from the wreckage? 769 770 One obvious lesson is to stay away from brand. Indeed 771 it's probably a good idea not just to avoid buying 772 brand, but to avoid selling it too. Sure, you might be 773 able to make money this way — though I bet it's harder 774 than it looks — but pushing people's brand buttons is 775 just not a good problem to work on, and it's hard to 776 do good work without a good problem. 777 778 The more subtle lesson is that fields have natural 779 rhythms that are beyond the power of individuals to 780 resist. Fields have golden ages and not so golden 781 ages, and you're much more likely to do good work in a 782 field that's on the way up. 783 784 Of course they don't call them golden ages as they're 785 happening. "Golden age" is a term people use later, 786 after they're over. That doesn't mean that golden ages 787 aren't real, but rather that their participants take 788 them for granted at the time. They don't know how good 789 they have it. But while it's usually a mistake to take 790 one's good fortune for granted, it's not in this case. 791 What a golden age feels like, at the time, is just 792 that smart people are working hard on interesting 793 problems and getting results. It would be overfitting 794 to optimize for more than that. 795 796 In fact there's a single principle that will both save 797 you from working on things like brand, and also 798 automatically find golden ages for you. Follow the 799 problems. 800 801 The way to find golden ages is not to go looking for 802 them. The way to find them — the way almost all their 803 participants have found them historically — is by 804 following interesting problems. If you're smart and 805 ambitious and honest with yourself, there's no better 806 guide than your taste in problems. Go where 807 interesting problems are, and you'll probably find 808 that other smart and ambitious people have turned up 809 there too. And later they'll look back on what you did 810 together and call it a golden age. 811 812 813 814 815 816 817 818 819 820 Notes 821 822 [1] The Bretton Woods agreement didn't fix exchange 823 rates between currencies directly. It fixed each 824 relative to gold. Obviously this also fixed them 825 relative to one another. 826 827 [2] The Golden Ellipse isn't quite a round rect, 828 because the sides aren't quite flat. It's similar in 829 shape to the superellipses popularized by Piet Hein in 830 the early 1960s, and in fact that may be where they 831 got the name. But mathematically it's not an actual 832 superellipse. My guess is that Patek's designer just 833 experimented with French curves till he got something 834 he liked. And to be fair it is a good shape. 835 836 [3] It was ironic that Patek Philippe of all companies 837 made this mistake, because Adrien Philippe was the 838 inventor of the modern crown. But they must have 839 realized what they'd done, because later Ellipses have 840 if anything excessively prominent crowns. 841 842 [4] The high ratio of design space to practitioners in 843 fine art has combined with the practical importance of 844 attribution to give people the impression that 845 painting in a distinctively Leonardesque way is what 846 makes Leonardo good. The most dangerous problem faced 847 by curators, art historians, and art dealers — the one 848 that has the worst consequences if they get the wrong 849 answer — is attribution. So inevitably they spend a 850 lot of time thinking and talking about the features 851 that distinguish the work of one artist from another. 852 But those aren't what make artists good. What makes 853 the line of a woman's cheek in a Leonardo drawing good 854 is how good it looks as the line of a cheek, not how 855 little it looks like lines made by other artists. 856 857 Because painting has such prestige, the myth that 858 having a distinctive style (rather than painting well) 859 is the defining quality of great artists has in turn 860 given cover to a lot of bad design in adjacent fields. 861 A brand that does something hideous to distinguish 862 their products can say "Like all great works of art, 863 ours have a distinctive style," and people will buy 864 it. 865 866 [5] An ad that Patek Philippe ran in America in 1970 867 famously described a Patek 3548 with a gold bracelet 868 as a "$1700 trust fund." Was it actually a good 869 investment? In the very best case a dealer might pay 870 you $20k now for one in unworn condition with its 871 original box and papers. That's about a 4.5% rate of 872 return, which is not absolutely terrible. But 873 apparently the average rate of return on S&P 500 874 stocks over this period was more like 10%, if you 875 reinvested all the dividends after paying taxes on 876 them. The average rate of return would have been over 877 9% if you merely bought a lump of gold that hadn't 878 been made into a watch. So, not surprisingly, the ad 879 wasn't very good investment advice. 880 881 [6] Tania Edwards, who ran US marketing for Patek 882 Philippe in the 90s, said that Bittel literally 883 sketched the design of the 3919 on a piece of paper. 884 This sounds odd to me, because the 3919 looked exactly 885 like the existing 3520 with the addition of sub 886 seconds (a small dial with a second hand above 6 887 o'clock). Why would you sketch a design almost 888 identical to an existing watch when you could just 889 point to the existing watch and say "that, with sub 890 seconds." What this story does show, though, is the 891 degree to which people within Patek felt their ad 892 agency was responsible for the design of the 3919. 893 894 [7] If I had to date the turning point for mechanical 895 watches precisely, I'd say 1986. Unit sales of Swiss 896 watches rebounded in 1985, but revenue didn't, which 897 means what we're seeing is the boom in cheap quartz 898 Swatches. Indeed, sales of mechanical watches must 899 have been down if revenue was flat despite the sale of 900 all those Swatches. Whereas in 1986 revenue turns 901 sharply upward even though unit sales only increase by 902 a little, which implies a corresponding increase in 903 sales of expensive mechanical watches. 904 905 [8] There is of course another reason some people are 906 into mechanical watches: because they're interested in 907 old technology. And if you are genuinely interested in 908 mechanical watches, there's good news. You don't have 909 to wear a billboard on your wrist or pay a lot to own 910 one. Just buy golden age watches. They still keep good 911 time, they're much more beautiful, and they cost a 912 fraction of what new watches cost. 913 914 The key to buying a golden age watch is to find a good 915 dealer, and the best way to recognize one is by how 916 much they tell you about the watch. A bad dealer will 917 just have a lot of fluff about the prestige of the 918 brand and the sleek lines of the case. A good dealer 919 will tell you the model number of the watch and 920 movement, have lots of pictures, including some with 921 the case back open, give you dimensions, disclose all 922 damage and restoration, and tell you exactly how 923 accurately the watch is running. Good dealers tend to 924 be watch nerds themselves, so they're into this kind 925 of thing. 926 927 (There are a few independent watchmakers trying 928 earnestly to make good mechanical watches now, but 929 their efforts show how hard it is to do good work when 930 the current is against you.) 931 932 [9] Oddly enough it might have helped that the 3919 933 was hand wound. If a watch runs for long enough, 5 934 seconds a day starts to add up. After three months a 935 watch that gains 5 seconds a day will be 7 minutes 936 fast. But with a hand wound watch you occasionally 937 forget to wind it, and it runs down. And when you wind 938 it again you reset it — on average to a time about 30 939 seconds behind the actual time. So if you forgot to 940 wind a 3919 every two weeks or so, it would rarely 941 have shown the wrong time. 942 943 [10] There's one brand still waiting to be reinflated: 944 Universal Genève, which was one of the big players of 945 the golden age but since 1977 has been little more 946 than a brand name passed from acquirer to acquirer. 947 They're scheduled to come back to life later this 948 year, no doubt with stories about their long tradition 949 of watchmaking. 950 951 [11] More precisely, a high ratio of size to accuracy 952 meant cheap. It's easier to make a larger movement 953 keep good time, but between two watches of the same 954 accuracy, the larger was usually the cheaper. 955 956 [12] Their form did once follow function. They were 957 originally diving watches. But they're long since 958 obsolete for this purpose. Present day diving watches 959 (now called dive computers) are digital and tell you 960 much more than the time. 961 962 [13] Rolex was awarded an average of 16.6 patents per 963 year in the 1950s, but only 1.7 per year in the 1960s. 964 965 Pierre-Yves Donzé, The Making of a Status Symbol: A 966 Business History of Rolex, Manchester University 967 Press, 2025. 968 969 [14] Rolexes also shared something more specific with 970 SUVs: aspirational manliness. An internal 1967 report 971 by Rolex's ad agency J. Walter Thompson explained the 972 idea they were trying to convey: "Because a Rolex is 973 designed for any situation, however rough or dangerous 974 or heroic or exalted, it implies that the man who 975 wears it is, potentially, a hero." 976 977 Reprinted in Donzé, op cit. 978 979 [15] This business model only works when purchase 980 decisions are driven mainly by brand. In a normal 981 market, if one manufacturer restricts production, 982 customers just buy from whichever competing 983 manufacturer offers something as good. It's only when 984 customers are seeking a certain brand rather than a 985 certain level of performance that you can manipulate 986 them by restricting its availability. 987 988 [16] Of course the first question one has on noticing 989 a bubble is: will it burst? The reason ordinary 990 bubbles eventually burst is that speculators get 991 overoptimistic, but in this case the CEO of Patek 992 Philippe controls the "money supply" and can thus take 993 measures to cool down an overheated market. So there 994 are probably only two things that could cause their 995 specific bubble to burst: if his successor is not as 996 capable, or if the whole custom of wearing mechanical 997 watches goes away. The latter seems the greater 998 danger. People aren't going to wear three things on 999 their wrists, so all it would take is for there to be 1000 two popular devices that were worn on the wrist, and 1001 mechanical watches would start to be seen by the next 1002 cohort of young rich people as an old guy thing. It's 1003 hard to imagine a luxury watch brand surviving that. 1004 1005 1006 1007 Thanks to Sam Altman, Bill Clerico, Daniel Gackle, 1008 Luis Garcia, the people at Goldammer, Jessica 1009 Livingston, Ben Miller, Robert Morris, John Reardon, 1010 D'Arcy Rice, Alex Tabarrok, and Garry Tan for reading 1011 drafts of this. 1012 1013 1014 1015 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1016 1017 References: 1018 1019 [2] https://paulgraham.com/index.html 1020 [3] https://paulgraham.com/brandage.html#f1n 1021 [4] https://goldammer.me/products/vacheron-constantin-18k-white-gold 1022 [5] https://collectability.com/learn/the-history-of-the-golden-ellipse/ 1023 [6] https://paulgraham.com/brandage.html#f2n 1024 [7] https://paulgraham.com/brandage.html#f3n 1025 [8] https://paulgraham.com/brandage.html#f4n 1026 [9] https://paulgraham.com/brandage.html#f5n 1027 [10] https://www.sothebys.com/en/buy/auction/2025/fine-watches-5/reference-5402st-jumbo-royal-oak-a-series-a 1028 [11] https://www.sothebys.com/en/buy/auction/2025/important-watches-2/reference-3700-11-nautilus-jumbo-a-stainless-steel 1029 [12] https://paulgraham.com/brandage.html#f6n 1030 [13] https://www.sothebys.com/en/buy/auction/2019/watches-online-6/patek-philippe-ref-3919j-a-yellow-gold-wristwatch/ 1031 [14] https://paulgraham.com/brandage.html#f7n 1032 [15] https://paulgraham.com/brandage.html#f8n 1033 [16] https://paulgraham.com/brandage.html#f9n 1034 [17] https://www.patek.com/en/collection/grand-complications/6300-400g-001 1035 [18] https://paulgraham.com/brandage.html#f10n 1036 [19] https://paulgraham.com/brandage.html#f11n 1037 [20] https://paulgraham.com/brandage.html#f12n 1038 [21] https://paulgraham.com/brandage.html#f13n 1039 [22] https://paulgraham.com/brandage.html#f14n 1040 [23] https://paulgraham.com/brandage.html#f15n 1041 [24] https://paulgraham.com/brandage.html#f16n