Chaos Monkeys: Obscene Fortune and Random Failure in Silicon Valley

by Antonio García Martínez

The trendy answer is that you need to find what’s called “product-market fit,” which is a fancy way of saying that you need to build a thing people are willing to pay for. It turns out that’s pretty hard, because you don’t know what people will pay for until you ask them to, and if what you’re building is truly novel, there’s no history for you to go on. Fortunately, the iteration cycle in a software business is fast (we don’t need to retool milling machines here), and from some approximately correct starting point we can converge on a final product, almost like successive guesses at the solution to an equation. The quicker we iterate, the more steps we can take in the direction of this mythical point of perfect fit.

Often at a tech company with an advertising business model bolted on to an otherwise user-facing product, the Ads team will be the weakest team in the company. Ads is perceived as a necessary evil, and marketing ain’t cool, so the hotshot twenty-two-year-old kids who are the recruiting cannon fodder don’t join up. The visionary CEO doesn’t care about money, only the user experience, and manages by looking at usage dashboards, not revenue ones. Bold product initiatives that will rekindle always-fickle app usage are green-lit, and their resources are allocated by CEO fiat. Meanwhile, like a nanny chasing a particularly petulant charge, the Ads team runs around reactively trying to monetize whatever comes out of the product-development process.

This other channel partly resembles Cold War drama, and partly resembles that time you announced to your fourth-grade class that you were going to kiss Becky Walker after sixth-period PE class, and all the resulting gossiping and cross talk. As events progress, the drama becomes more and more infantile, convincing you finally—as if you needed more proof this late in your startup trajectory—that humans, even at the rarefied heights of the economic elite, are in truth scared, needy children playing at dress-up and pretending to be grown-ups.

Is it a reasonable thing, I ask you, for a grown man to run about and hit a ball? Poker’s the only game fit for a grown man. Then your hand is against every man and every man’s hand is against yours. Team-work? Who ever made a fortune by team-work? There’s only one way to make a fortune and that’s to down the fellow who’s up against you. —W. Somerset Maugham, “Straight Flush”

The risk expectations of founders and investors can often be severely misaligned. I hate sports analogies, but here’s one to explain this vital point: most VCs are playing a version of baseball in which the only way to score is to hit a home run when you’re at bat. They don’t care if you disgrace or impoverish yourself and strike out, and they don’t care if you get a solid line drive that lands you on second. To them, strikeouts and getting on base are equally pointless, and so they’ll push to proverbially “swing for the fences” no matter the count or the team you’re up against. The reason for this all-or-nothing approach is how their funds are structured. VCs (or so-called angels like Sacca) raise a fund, out of which they’ll provision some number of investments. Barring doubling down on the same company, which they might do if the fund still has money when a company raises again, those investments are effectively “fire and forget.” The fund’s total profit will be calculated from whatever those initial bets return. Unlike, say, a hedge fund portfolio manager, who rolls the winnings from one good bet into the next, compounding a series of returns into something truly huge, VCs do not take liquidity from one company’s exit and pour it into yet another’s.* This, at heart, is why the go-big-or-go-home strategy makes the Silicon Valley world turn, and why entrepreneurs push themselves to be either the next Airbnb, or nothing. The entrepreneur who bucks this and creates a long-...

If you can only be good at one thing, be good at lying . . . because if you’re good at lying, you’re good at everything. —@gselevator, July 25, 2013 FRIDAY, APRIL 1, 2011

As with any interview, whether it be a hardball technical/engineering interview, or the foofy product manager one, the challenge is very simple: either you’re incompetent and you struggle to provide even one answer you hope is right to the interview question, or you’re not, and your brain produces two or three. Each of those could be right, depending on your interviewer. Truth in the world resides only in mathematical proofs and physics labs. Everywhere else, it’s really a matter of opinion, and if it manages to become group opinion, it’s undeservedly crowned as capital-T Truth. And so you need to determine whatever the local version of truth is you’re inhabiting. And that you do by reading body language, and understanding the intellectual sauce your opponent has been marinating in (if any). With that in mind, you choose among the three alternative answers you could spout.

Here’s another truth about tech life: anyone who claims the Valley is meritocratic is someone who has profited vastly from it via nonmeritocratic means like happenstance, membership in a privileged cohort, or some concealed act of absolute skulduggery. Since fortune had never been on my side, and I had no privileged cohort to fall back on, skulduggery it would have to be.

Faster, faster, faster until the thrill of speed overcomes the fear of death. —attributed to Hunter S. Thompson

In a posthistorical developed world devoid of transcendent values, whose pantheons look like North Korean grocery stores, bare shelves empty of any gods or heroes, this corporate fascism was intoxicating.

Interesting take on why ppl join trch companies - Because the modern world has lost meaning as belief in religion, mythicism has waned

Sturmabteilung.

Cynicism is the last refuge of the shiftless.

Nice quote

The human need for immortality projects—those ends that dole out meaning and purpose beyond ourselves—hasn’t changed since the pyramids. The only difference now is the nature of the putative Holy Land, and the means for achieving it.

Whatever the flavor of title, what does a product manager, by whatever name, actually do? The MBA-esque job description would be “CEO of the product,” because those B-school pukes like sporting acronymic titles. This is many a company’s definition of the role, and it’s not completely wrong, though it makes the job seem statelier than it is. A more illustrative description is “shit umbrella.” If you imagine a bursting monsoon of clumpy diarrhea pouring down like God’s own biblical vengeance, that is more or less where you find yourself in either a startup or a large, high-profile, and complex organization like Facebook. You, my dear product manager, are the communal manservant to your engineering team, holding a large, cumbersome paramerde above their heads bent over keyboards on which they furiously type. By definition, what you do is everything that needs to be done, other than hands-on-keyboard type code. So that means sitting in endless meetings with the privacy legal team, giving highly selective and edited versions of what your product is going to do, and explaining how it fits into some antediluvian legal rubric. It means pitching a roomful of smiling and empty-headed salespeople, so they can start priming the client pump and bring spend into your new product baby. It means wheeling and dealing with other PMs to either cajole a product change or cadge some engineering resource. It means fronting for the product in high-level meetings with senior management, and tryin...

Nice description of what a product manager does in a texh company

The Ads team’s mandate was very simple: make more money, but don’t piss off users while doing so. The monetization side of Facebook was still stuck in that subservient role typical of earlyish-stage startups that are mostly consumer-facing, and with a CEO who abhors grubby lucre. Kind of like a patient mother trailing her destructive toddler and picking up discarded toys along the way, Ads was expected to make money with whatever new features the user-facing side of the company launched.

Any culture able to shut itself off from the outside world goes insane in its own unique way, and Facebook had essentially done that with its Ads team.

apotheosis

Pelayo

my Sisyphean task

consigliera

Every quarter Sheryl scheduled a gargantuan meeting meant to show off the wonderful tools engineering was building for Sales, and how well the hybrid engineering-operations teams were collaborating. Sheryl’s managerial prowess was on full display in these powwows, as she skillfully held court among the assembled lieutenants of the various subrealms. Picking up subtle psychological cues from an off-the-cuff remark here, unearthing some lingering issue that lay dormant over there, making sure every voice was heard but no voice heard too much, tamping down some burst of irrelevant drama to keep the action moving, the woman knew how to run a roomful of big names and even bigger egos.

expectorated.

If this stretch of Facebook life seems chaotic in its random divagations, that’s because it was random and unpredictable. Life as a Facebook product manager was less being a lieutenant in some stolid industry juggernaut, and more being buffeted by constant external and internal forces. The external forces were the ever-present machinations of Google, the general temperature of the advertising world with respect to Facebook’s marketing offerings, and the fortunes of Facebook’s social media vision (e.g., were teenagers still using it?). The internal forces were the fluctuations in your standing in the impromptu hierarchy of people and products inside Gokul-land, as well as the fate of the products you managed and nursed to success (or at least to quick failure). Particularly on my beat, we were constantly sprinting half-blind through a minefield of potential legal problems, and encouraged to do so as the risky cost of innovation. For my entire career at Facebook, I was embroiled in a rolling debate with the Facebook privacy and legal teams about what we could and couldn’t get away with, chiseling away at their legal trepidation, and trying to find some legal rubric that would forgive (or at least defensibly excuse) our next depredation with user data. One day, though, and with little prompting on our part, that all came to one of those do-or-die confrontations, of which Facebook had many in its bustling drive toward Internet domination.

Feels like Uber

kvetch

Privacy is to Facebook as nuclear weapons are to Iran: this constant cloud overhead, obsessed over by outsiders even if poorly understood, and the starting point of pretty much any conversation with an outside player.

detritus

rebus

This was Facebook culture for you: lots of bold, unconventional experiments, mostly failures with some notable successes, an immediate course correction to prune the failure, and then internalizing the experience via the culture. The crap murals and bluetape were as core to Facebook as the Like button (and Beacon).*

But the first rule of startups is also true of any fast-paced, competitive workplace like Facebook: act like you belong there, even if you don’t.

lapidary

As the ever-sagacious @gselevator quoted: in communism people made lines for bread, while in capitalism they make lines for iPhones. Sure, iPhones are better than bread, and the standard of living in capitalist countries is clearly higher, but the lived experiences of either, from the point of view of the working proles, bears more than a passing resemblance.

In 1956, the sociologist Leon Festinger published a landmark study of a cult formed around a Chicago housewife named Dorothy Martin. Martin channeled messages from extraterrestrial beings living on different “vibrational planes,” which she recorded via automatic writing. Her messages predicted a catastrophic flood that would destroy the United States on December 21, 1954. The acolytes of the cult’s prophet would be saved via alien spacecraft, which would whisk them off to the higher planes of existence they had been specially selected to experience. The cult’s membership grew in time, and as the apocalyptic date approached, the members left jobs, let properties and businesses languish, and alienated their disbelieving families in expectation of the end times. When the flying saucers and ensuing apocalypse failed to appear on the appointed date, the cult’s believers did not lose faith. On the contrary, the experience bolstered their beliefs, annealing them into an intimate confederacy of false belief. Vestiges of the cult persist even to this day. This study would lay the groundwork for Festinger’s theory of “cognitive dissonance”: the mental stress people suffer when presented with realities contrary to their deeply held beliefs. The key takeaway is that humans naturally avoid this discomfort, skirting situations that aggravate it, or ignoring data that make their mental contradiction more apparent. Note: The purpose of the following exposition is not a neener-neener troll...

A good history on how the term “cognitive dissonance” was coined

The advertising agencies, those moderators of commercial taste who lived in New York and were paid lots of money to spend lots of money, convinced prominent brands to buy Likes for their Facebook pages. Companies like Starbucks (on whose board Sheryl sat) or Burberry (they of the $4,000 cashmere coats) routinely forked over $10 million in order to gain five million Likes for their page. Facebook salespeople would preach the miraculous power of Likes the way a Catholic priest might sermonize on the eucharistic miracle, bread and wine turning into body and blood. However, it was never quite explained how Likes would transubstantiate into actual sales dollars. This was one of the mysteries of the Facebook faith that believers had to simply accept as beyond human understanding. To attempt to measure any of this, as with radiocarbon dating the Shroud of Turin, would have been spiritual buzzkill and fucking with the magic, so shush and pay.

In early days - Facebook’s primary monetization strategy was “buying likes”

It involved a nebulously titled product called Open Graph, the first version of which had launched at Facebook’s developer conference F8 the year before I joined.* A later version added the verbal dictionary that accompanied “Like,” expanding the Facebook vocabulary to things like “play,” “listen,” “watch,” or “buy.” It was the new subject-verb-object language of everything you did online. “Antonio García Martínez listened to Wax Tailor’s ‘Only Once’ on Spotify.” Rather than merely express some vague approval via Like, Facebook users could now broadcast everything they were doing, with the aid of outside developers who built Facebook’s new grammar into their products. By doing so, these developers made their products “social,” and potentially viral. In exchange for pumping their data full-throttle into Facebook, those outside developers—music players like Spotify, or publishers like the Washington Post—got News Feed distribution, driving yet more users to their content and services. That was the dream, anyhow. The monetization side of this was almost certainly not what you’re thinking. Mining or selling the data wasn’t the point. By oblique analogy to Google’s AdWords, in which promoted results appear alongside regular search results, the distribution of certain Open Graph stories would be “boosted” and appear more often in a user’s feed. Zynga and Eminem would in theory pay to have their stories boosted, and have you, the user, engage with their product or content more of...

“Sponsored Stories” (Open Graph) - Another monetization product that failes

Sponsored Stories was the classic indefensible two-miracle startup idea. Miracle one was all the companies in the world doing work to integrate with Facebook and give away their data, at dubious benefit to themselves. Miracle two was those companies’ marketers abandoning their old workflows and success myths for some newfangled paradigm, one with unproved and unprovable efficacy, just because Facebook said so. And so, like any two-miracle startup idea, it was almost certain to fail.

Grouped: How Small Groups of Friends Are the Key to Influence on the Social Web.

According to Facebook Research, the inclusion of social context (i.e., the addition of “your friend Joe liked” to a text ad) lifted clickthrough rates something like 40 to 60 percent versus similar ads without your friend’s smiling face. That sounds like a lot, and certainly it was better than the alternative of no effect. In practice, though, this was terrible news. The clickthrough rates on Facebook ads were abysmal, mostly due to their crappy targeting. A 0.05 percent clickthrough rate was average; achieve a rate of 0.11 percent and you were ordering the truffled steak at Alexander’s in SoMa (on the client’s dime, of course). Compared with regular display advertising, Facebook ads didn’t even register; even the worst-targeted ads (e.g., those moronic “punch the monkey” pseudogame banners) got clickthrough rates of at least 0.1 percent, and well-targeted ads using first-party data and rich, dynamic creative got click rates as high as 1 percent or more. It meant Facebook ads underperformed by as much as twenty times, if you were actually doing the bookkeeping (which many FB advertisers were not). Thus, a 60 percent bump in performance was nice, but it was piddly-boo compared with the lift generated by real, piping-hot targeting data. Facebook had bet its entire monetization future on a scheme that modestly boosted clickthrough, and even then it wasn’t clear why; the Facebook Research paper mentioned a host of possible experimental confounds. I had seen an early version of...

misanthropy,

Among the various weird products I happened to manage while at Facebook, one is particularly relevant. Let’s flash-forward for a moment. The Logout Experience (LOX) was one of Facebook’s mid-2012 IPO period “completely fucking desperate let’s make more money now” products. Basically it was this: when you logged out of Facebook, instead of seeing the usual Facebook log-in/register page, you would see what amounted to the upper half of a regular Facebook page; that is, the cover photo and header that we all know from our own Facebook profiles. The idea was to sell this space to the richest and stupidest advertiser of all, the legacy-admissions student of digital media: the brand marketer. Brand marketers, like politicians, are people paid to spend other people’s money. In this case, their putative goal is to raise “brand awareness,” that diaphanous and quasi-intangible substance that makes you covet a $10,000 Rolex Submariner when you finally get that bonus or promotion. The other equally trigger-happy marketer who would supposedly devour LOX was the Hollywood-centric agency that wished to trumpet the release of another cinematic crime against humanity. The Logout Experience was Facebook’s attempt to present a tempting target to those marketers, the equivalent of a large banner ad or takeover. The name “LOX,” a bit of product marketing genius from Scott Shapiro, stemmed from the historical tradition of naming Facebook Ads products after either birds or fishes. Given this was...

“Logout Experience (LOX)” - Another facebook monetization produc

Modern advertising is just selective name-calling, which is why it’s so wonderfully primal. How is that? The only difference between the various marketing channels we think are so distinct is the names we use to address the target audience. How advertising works is effectively a call-and-response of names, with some mechanisms being far more efficient than others. So what are those names? In the direct-mail advertising world, it’s the postal address we affix to a piece of third-class mail. For example, if Bed Bath & Beyond wants to get my attention with one of its wonderful 20 percent off coupons, it calls out: Antonio García Martínez 1 Clarence Place #13 San Francisco, CA 94107 If it wants to reach me on my mobile device, my name there is: 38400000-8cf0-11bd-b23e-10b96e40000d That’s my quasi-immutable device ID, broadcast hundreds of times a day on mobile ad exchanges. On my laptop, my name is this: 07J6yJPMB9juT0war.AWXGQnGPA1MCmThgb9wN4vL0Upg.BUUtWg.rg.FTN.0.AWUxZtUf This is the contents of the Facebook retargeting cookie, which is used to target ads to you based on your Web browsing. Though it may not be obvious, each of these keys is associated with a wealth of our personal behavior data: every website we’ve been to, many things we’ve bought in physical stores, and every app we’ve used and what we did there.* The longevity of these keys and their associated data is important, however. We move our physical address less often than we cle...

Mobile - Desktop - Offline identity How do we combine these?

I submit that the role of modern-day Big Brother is actually played by companies you’ve likely never heard about. Companies with names like Axciom, Experian, Epsilon, Merkle, and Neustar (among others). These are the companies that since the dawn of the direct-marketing age in the sixties and seventies have been tracking all of consumer America. They know your name, address, phone number, email address, education level, rough income, who else is in your household and their ages and spending patterns, and which consumer segment you fit into—and they’ve been accumulating this information since before the Internet even existed.

And how did this enormous nationwide surveillance apparatus, without even the merest hint of oversight or regulation, come to exist? The mail, ladies and gentlemen. That’s right, the postage-stamp people. Get this. By the Post Office’s own figures, direct-mail advertising resulted in over $17 billion in revenue to the Post Office in postage alone, propping up what would otherwise be a bankrupt organization. The entire direct-mail industry, that crap in the mail you throw away, is surely north of a $50 billion a year industry if you consider design, printing, targeting data, and postage. To put that in perspective, all of Google makes just over that much in a year. In 2014, online marketers spent about $19 billion on display (basically, the Internet other than Google and Facebook). So we’ve got almost another Google, three Internets, or three Facebooks of money waiting in the offline sidelines. Think about that for a second, particularly if you, like me, live in the maelstrom of pixels and electrons that characterizes contemporary digital life. To do direct mail, you have to kill a tree, make it into paper, hire a designer, print something pretty on it, properly package or envelope the thing, and then pay the Post Office around $.20 in bulk to send it. Do the math. It works out to about $1,000 CPMs (cost per thousand views). Most online media sells for somewhere around $1 CPM, perhaps a bit more. That means online advertising can get away with being a thousand times less ef...

How online identity gets matched to offline identity by companies like Datalogix, Neustar, Liveramp

On-boarding in the Facebook context is even cleverer. Facebook and companies like Acxiom and Datalogix have compared personal data (with none sharing actual data with the other, again via the miracle of hashing), and joined the universal FB user ID to the analogous IDs inside Acxiom, Datalogix, and Epsilon. The advantage that Facebook and Google have over the regular data on-boarders is twofold: they have much more of your personal data, and they see you online all the time. The match rate (i.e., the percentage of offline personas that can be found online) for Facebook’s Custom Audiences product, introduced in 2012 (and more on that shortly), is as high as 90 percent. That means for every hundred people marketers target via Custom Audiences, Facebook will find ninety of them, a shockingly high fraction in the fuzzy world of advertising. Facebook, Google, and others have achieved the holy grail of all marketers: a high-fidelity, persistent, and immutable pseudonym for every consumer online. Even better, they’ve joined that to your real-world persona, the one that shows up bleary-eyed at two a.m. at a Target in El Cerrito looking for tampons or a six-pack of Natural Light. Incidentally, this is all public, and fully documented by Facebook press releases. It’s just poorly understood and no one thinks about it. But people probably should.

He represented the best in Facebook engineering: irreverence without disrespect, competence without arrogance, ambition without ego.

Seen from the jaundiced eye of the outside advertising world, Facebook had for years been completely unconnected to any outside data source, with no real tracking or attribution to figure out which Facebook ads were working (and where). Now we were proposing rendering a quarter of the Internet targetable via all the online data in existence. Every product you’d looked at or bought, everything known about your online reading or browsing habits—it was all going to be inside your Facebook experience now. There were two ways to do this. The first would end up being called Custom Audiences (CA), and was an extension of the existing ads system. Like the data “on-boarding” we talked about earlier, it would be a way to join offline data like names, addresses, and emails to Facebook users. That was its original and sole intent, though it soon grew into much more. The second, more interesting approach, Facebook Exchange, or FBX, would form the final chapter in the programmatic takeover of digital media: a real-time exchange connecting the world’s advertising data, user by user, and ad impression by ad impression, to the Facebook experience.* How did it all work? In the CA case, advertisers had to upload lists of email addresses, phone numbers, names, or other personally identifiable information (PII), and manually craft some targeting cluster (“people who bought something last month”). The Facebook technology, drafting as it was off recycled targeting technology, was creaky and slow...

“Custom Audiences” and “FBX” - Another Facebook monetization tools

Zach Coelius

FBX wasn’t merely a new ads feature. It was an entirely new ads system parallel to the existing Facebook one. The exchange would have separate data stores with billions of lines of data, reflecting the various data joins to the outside world we had made. The FBX code itself, unlike anything inside Facebook Ads, would have to contact machines all over the world for bids and ads in real time, and react to bids just as quickly, all of this tens of billions of times per day, hundreds of thousands of times per second. How ads themselves were created and uploaded would be different, allowing advertisers that much-coveted (and before FBX impossible) ability to dynamically fine-tune ads for every user and every ad experience. Ads statistics, the boring bean-counting task critical to keep the budgets flowing, would have to exist separately and work for advertisers and their accounting departments. Not to mention a whole suite of internal tools and dashboards to monitor the health of this beast we had created, making sure bids weren’t getting dropped, or we weren’t slamming our partners with too many requests for bids. On the business side, we (by which I mean the royal PM “we”) would have to find advertising software companies as partners, and on-board them with a list of documents, specifications, and integrations tests.

What’s my big beef with capitalism? That it desacralizes everything, robs the world of wonder, and leaves it as nothing more than a vulgar market. The fastest way to cheapen anything—be it a woman, a favor, or a work of art—is to put a price tag on it. And that’s what capitalism is, a busy greengrocer going through his store with a price-sticker machine—ka-CHUNK! ka-CHUNK!—$4.10 for eggs, $5 for coffee at Sightglass, $5,000 per month for a run-down one-bedroom in the Mission. Think I’m exaggerating? Stop and think for a moment what this whole IPO ritual was about. For the first time, Facebook shares would have a public price. For all the pageantry and cheering, this was Mr. Market coming along with his price-sticker machine and—ka-CHUNK!—putting one on Facebook for $38 per share. And everyone was ecstatic about it. It was one of the highlights of the technology industry, and one of the “once in a lifetime” moments of our age. In pre-postmodern times, only a divine ritual of ancient origin, victory in war, or the direct experience of meaningful culture via shared songs, dances, or art would cause anybody such revelry. Now we’re driven to ecstasies of delirium because we have a price tag, and our life’s labors are validated by the fact it does. That’s the smoldering ambition of every entrepreneur: to one day create an organization that society deems worthy of a price tag. These are the only real values we have left in the twilight of history, the tired dead end of liberal de...

jejune

ab initio