Remains of the Day: Issue 04

The most addictive multi-player game/debt in tech/zealots only/user pacing as design element/movie of the week

As a reminder, I’m Eugene Wei, a former product person at companies like Amazon, Hulu, Flipboard, and Oculus, and I’ve kept a personal blog Remains of the Day since 2001, offering what I describe as “Bespoke observations, 80% fat-free”, though the perceived fat ratio will vary depending on your tastes in technology, media, and all the other random topics I write about. Also, according to some doctors, fat isn’t as bad for you as once thought, though this does not constitute medical advice.


We Now Resume Regularly Schedule Programming

I spent the first month of 2020 dealing with water damage to my condo. Lots of calls to insurance companies, then contractors, then back to insurance companies. The endless volley of estimates and invoices. Then, finally, finding a moving company to move everything touching the ground out of my condo so a new hardwood floor could be installed.

As of now I’m still unpacking, and my condo is still a mess of things that need to be put back in place, but at least I’m not confined to my bedroom. Homeownership may be encouraged by the tax code in the U.S., but the only form which seems enticing to me is the one in which I am wealthy enough to build a home from scratch, an act of creativity.

As I emerged from the depths of the water damage incident, I was struck down by the flu, which seems to be making the rounds in the Bay Area this season. For a week I oscillated between a fever and chills that left me shaking even beneath several blankets. My doctor friends told me it would have been worse had I not gotten the flu shot, but that proved little consolation in the moment.

At any rate, back to a more normal writing schedule, I hope.


The Largest, Most Addictive Multi-Player Game in the World

I just finished Ezra Klein’s Why We’re Polarized, a thought-provoking look at the structural mechanisms driving the hyperpolarization of American politics. One popular culprit that he largely avoids is social media; as he notes, the polarization began long before social media.

However, I do think that the internet is a massive amplifier of the ideological sorting that Klein identifies as one of the key causes. More than a few times, as I read the book, I thought to myself that the subtitle of the book should be Group Status as a Service. Klein notes that our identities now stack and multiply along such clean lines between the two dominant political parties in a way that wasn’t true in previous eras of American politics, like the so-called political golden era from just after WWII through the passage of the Civil Rights Act.

I read Clay Shirky’s Here Comes Everybody: The Power of Organizing Without Organizations years ago, and it was, at the time, a very positive look at how the internet enabled global coordination in a way that wasn’t possible before. The discovery of people with shared interests from across the world rather than just your local geographic circle is indeed one of the joys of the internet and continues to be.

At the same time, in the same way that gaming inevitably shifted towards multi-player synchronous modes as the internet connected the world, it almost seems inevitable in hindsight that the internet would converge on identity-based media as hundreds of companies experimented on the most addictive formats, driven by growth teams and machine learning. Identity anything is a hell of a drug for a social species.

When I’m asked about what the most compelling media developments are today, I often point to various innovations in gaming; we live in the post-Fortnite era of gaming and we aren’t ever going back. But I always add the following: the most addictive, multi-platform, multi-player game in the world today is not Fortnite but identity politics. You can play it on any social media site, regardless of device or operating system, in almost any media type and in a variety of modes from single to multi-player, from battle royale to coop.

Furthermore, identity politics is, like Facebook Connect, a platform that integrates into anything. When Donald Trump trolled everyone by complaining about Parasite winning the Oscar for best picture, he was bringing identity politics to film. But even if he hadn’t, it was already there. Identity politics is a human web service that has been integrated into every adjacent field more seamlessly than at any point in human history.

Like kale? Sip lattes? Or perhaps you’re a voracious red meat eater? All of that maps to some identity. Do you listen to Kendrick Lamar or Kanye West? Did you prefer Moonlight or La La Land? Do you drink wine or beer? If you drink beer, are you drinking Budweiser or some microbrew? There is a legend that maps all of those to some identity.

You may decry it as destructive and emotionally taxing, but we’ll never really grapple with identity-based warfare unless we acknowledge its appeals. I didn’t jump in the other day post Democratic debate as I try to avoid the ineffectual political posturing on social media as much as possible, but I’d be a liar if I didn’t confess to deriving some pleasure from the absolute merciless social media skewering of Bloomberg’s debate performance on social media.

The problem of this multi-player identity politics game is that the choose-your-character screen offers only a few choices. The advantage of selecting one of the stock characters is that they immediately put you into a massive tribe, along with all the identity-based network effects that brings. The problem is that the stock character choices are already pitted against each other in challenge mode.

If you could’ve invested in identity-everything about 10-15 years ago it would’ve been a surefire 100X return. Investing in social media turned out to be a good proxy, and Trump rode his all-in bet to the White House.


Once Upon a Time in Tech

Alex Danco wrote another great piece on why debt is coming to tech. As I was reading it, I realized that it wasn’t just a story about what is coming. It was a story I already lived once.

I began my career at Amazon in strategic planning, maintaining a large spreadsheet model of the business. Strategic planning is the forward-looking counterpart to accounting, and so my model projected what was going to happen in the quarters to come. This would help us manage our relationship to the Street and to raise money when needed.

The income statement model had two key components, the revenue side and the cost side. I built up the cost side department by department; some were purely variable, others were semi-variable (after removing all variable costs we had gross margin, which we reported to the street, and after removing semi-variable costs as well we had direct margin, which we kept to ourselves), and the rest were fixed.

Writes Danco:

Where could you put debt to work effectively? Oh, I dunno, how about that thing that every tech company does now: creating customers! You have to spend a bunch of money today to acquire users. But once you have them, they send back recurring revenue that’s pretty predictable at a cohort level. Hmm! Tech companies with recurring revenue business models are this close to connecting the dots.

Guess what Amazon’s revenue model for books, music, and video, our first three retail businesses, resembled: recurring revenue of utter predictability at the cohort level!

In practice, our forward-looking revenue model simply layered the revenue from every previously acquired cohort of customers (we called each cohort a vintage, defined as the group of customers who made their first-ever Amazon order in a specific month and year) beneath the projected revenue from the new customers we’d acquire in any future month. It turns out that at that time, when media came predominantly in physical form, people buy books, CDs, and DVDs at a startlingly predictable frequency interval when grouped into sufficiently large cohorts.

Our revenue model was accurate within 2-3% month after month, and if you go back to those early quarterly calls with investors you’d see that Amazon beat projections every. Single. Quarter. This was because we could set our guidance for the street below our projected revenue with almost perfect precision. It was just a matter of dialing in how much of a beat Jeff and Joy (our CFO) wanted for a particular quarter. I’m almost certain I’ll never work on a more accurate spreadsheet model in my life again; when Amazon started entering low purchase frequency, high ticket price retail categories like refrigerators and plasma televisions the model’s accuracy surely dipped considerably.

As it happens, the last major financing I worked on for Amazon was what was one of the largest convertible debt offering in history up until that moment. We didn’t turn to the stock market for another round of equity financing. With a business as predictable as ours, it made no sense to dilute.

Those who were in tech at the time will recall reading articles about Amazon with sensational headlines like Amazon.bomb and Amazon.toast. You’d think, scanning the press coverage of Amazon, that it was a volatile speculative stock. The truth, as I could see every time I stared at the spreadsheets, was that Amazon was an almost boringly predictable recurring revenue business. My spreadsheet model looked like a model of a portfolio of annuities.

[Seeing how predictably people purchased books, music, and video gave me a sense of one slice of Maslow’s hierarchy as being a predictable narrative quota which people the world over try to satisfy each day, but that’s a topic for another issue.]

Of course, to get Amazon off the ground, Jeff Bezos did give away some equity. Given how new the internet was at the time, that was understandable. But had you been able to time travel to the future and show someone my spreadsheet from 1998 with the name of the company redacted, you would be hard-pressed to identify it as a risky internet stock. With its negative operating cycle, it would look more like an insurance company than a retail company.

All of this is just to say that what Danco writes about isn’t just coming, it already came. Once upon a time.


No Country for Non-Zealots

Something else Danco writes about in his piece is that area where Financial Capital (FK) and Production Capital (PK) diverge during the Installation Period per Carlota Perez’s classic text Technological Revolutions and Financial Capital. In her model, in the installation period, FK and PK diverge in a zone of speculation where VC’s are hopeful of explosive returns but the variance in outcomes is still massive.

The categories where FK and PK are most in tension right now are the four horsemen of internet hype: VR/AR, AI, self-driving cars, and cryptocurrency. In all four, I’m very bullish in the long run about their enormous impact on the world.

But when I say “long run” I mean really long run. I suspect we are going into a second or third winter (depending on your sense of how seriously each of those had a spring in the past) for all of them, and these cold spells could last past long past the liquidity timing expectations of a lot of investors.

As an employee working in one of those fields, that may or may not be relevant. Where it does matter is in your life liquidity preferences. That is, if you’re going to work in one of those fields, you need to determine whether you’re in it because you can’t fathom working on another problem or because you think it’s the next hot wave to ride onto a tropical beach of prosperity sometime soon. If it’s the latter, check your emergency reserves of patience and dedication and see if they’ll survive a pandemic of pessimism and capital.

Naval once told me that you should only work on crypto if you’re a zealot. Because all the people you’re competing with will be zealots. Ten, fifteen, twenty years of your career...that’s nothing to a zealot. If you’re not, the opportunity cost is a lot higher.

If you are a zealot, great! Winters wipe out those who are weak of constitution, low on faith.


Rhythm is a Dancer

The difference in UX of various forms of media can seem subtle yet make all the difference. I was trying to explain to someone the shift from PowerPoints to written memos at Amazon, the feeling of that switch. Yes, part of is that the medium of long-form prose is less forgiving of throwing a spaghetti of ideas against the wall in bullet point form. Prose is, if nothing else, a filter for lousy argument structure.

But underrated is just how much it mattered to Bezos to be able to consume the content at his own pace. In the lead-up to the switch from Powerpoint to memos, I sat in many a meeting where someone would be going through a slide deck for Jeff and he had already skipped ahead in the printed copy to some much later slide. That pacing problem would visibly frustrate him.

As a product manager or designer, one axis of consideration is the scope of user control in the pace of consumption. One of the underrated differences between Stories and Feeds as products is that the method by which users control the pace of each is different. You can scroll a feed with your thumb, but each story still consumes some fixed amount of vertical pixels that require some fixed amount of scrolling to bypass. In Stories, by contrast, you can tap as early as you want to skip a Story and advance to the next. There’s a sort of rhythmic vigor and sense of achievement that comes with zapping an entire story off the screen at your own discretion.

If you’ve ever attended an Amazon meeting where everyone was quietly reading a memo for the first ten to fifteen minutes, you realize one of the benefits is allowing every attendee to process at their own pace. The old method, where everyone must succumb to the speed of the presenter’s delivery is a sort of storytelling tyranny. If that presenter is Steve Jobs, it works because he’s Steve f-ing Jobs. In most other cases it is maddening and a suboptimal allocation of the attention resources of a large room.

Even between two forms of presentation that both allow a user to control the pace, the feeling of consumption can differ depending on the interface. Reading one of Robin Sloan’s tap essays differs from reading that same essay as one single block of text. The former feels like a brisk walk, the latter more like rappelling.

Jenny Offill’s new novel Weather presents sentences as stand-alone, each set off with lots of white space. Every sentence almost looks like a tweet or a proverb, and I find myself dwelling on each with extra care and scrutiny. It’s not a long novel but reading it feels like a series of deep breaths in meditation than the high-intensity saccadic intervals I use to mow through long non-fiction books.

You can grant a reader just a smidge more agency and that changes the experience entirely.

Setting a sentence by itself, like this, is like a speed bump to the reader’s reading velocity.

I’d love to use an app that allows one to construct tap stories, but I’d prefer that app not be Keynote/Powerpoint or Snapchat/Instagram. Sometimes I think about making an app that helps you make faux silent movies, using speech-to-text to create the dialogue cards that follow each movie clip. That will probably end up as a TikTok filter in due time.

Speeding up podcasts is another example of user-controlled pacing. Sometimes just that little bit of interactivity is enough to transform the experience. Spotify should add that feature for podcasts in the future.

Books can be read at one’s own pace while video can’t be scanned easily, often cited as a reason infovores prefer text to video online. But now that video playback speeds can be altered this may change the relative balance of appeal of those mediums to people. Netflix debated adding video playback speed controls to their streaming app and film lovers freaked out. Creators of more traditional media types like books, music, and film will always tend to prefer to retain control over presentation as much as possible. A more modern storyteller understands that delivering media via interactive devices means that pace is now another option in the storytelling palette.


Movie of the week

My movie pick of the week is a Korean film dealing with the deep, long-simmering, and widening division between the economic classes within Korean society, a tension that boils over in the film’s bloody final act. A film which taps into deep veins of resentment over economic inequality crisscrossing the world.

I’m talking not of Parasite, but of Burning, a film that came out a year before Bong Joon Ho’s Oscar-winning picture. Burning is one of the best movies of the century, and in many ways, it is Parasite’s spiritual predecessor.

If you’ve read the short story “Barn Burning” by Haruki Murakami on which the film is based, or any Murakami, you’ll understand why adapting his work to film is such a challenge. The interiority of his protagonists and his unique dialect of magical realism are often the ingredients for an overly twee yet still opaque arthouse cipher. Like many, I considered Murakami’s works to be unadaptable to film (this year, Denis Villeneuve tackles another white whale film adaptation project, Frank Herbert’s science fiction classic Dune; normally I’d say that David Lynch might be the only type of director suited to directing a Murakami text but he already took a shot at directing Dune and it flopped).

I was wrong. It turns out a good Murakami adaption just needed the right director. I loved Lee Chang Dong’s two previous films, Secret Sunshine and Poetry, but he hadn’t made anything in eight years and his absence from directing had started to assume a mystical quality, like Terence Malick’s even longer hiatus after directing Days of Heaven. When I saw Lee was making his return at long last with Burning, and after having heard some early critical buzz about the film from Cannes, I put it at the top of my list for the Toronto Film Festival.

Burning made something like $7 million in global box office; Parasite has pulled in over $200 million globally. The ratio reflects, in rough doses, their relative efforts to appeal to a mainstream audience, viewer be warned. Burning works at a lower level of abstraction, half-submerged in the subconscious. Many scenes feel like a dream; while watching them you feel a visceral logic that starts to slip through your understanding as soon as you try to explain it to yourself. If you like your movies tidy, this isn’t going to be your jam.

Where American Psycho is aggressive with its social satire, swinging axes at you both literally and figuratively, Burning teases. Is it a serial killer film? A murder mystery? A missing woman film in the grand tradition of Vertigo? A Biblical tale of fratricide? A Korean Great Gatsby? All of the above, and none of them.

Some movies are rewatchable for the way they execute the expected; as with pop music, we know what’s coming but the arrival of the familiar is the reward. We know where the bass drops and it’s still satisfying the umpteenth time it does.

Burning is a film that’s rewards repeated viewing with persistent unease. I’ve seen the film several times, and it always leaves me feeling one step closer to some truth and yet as far away as ever from grasping what that is. The chase is the thing.

Actor Steve Yeun yawns at one point in the film. It’s likely the most important yawn in film history, holding the totality of the indifference of the wealthy in one gesture. I’m haunted by it still.

[Burning is currently streaming on Netflix.]