Failures happen when you're not smart enough to actually deliver results in your preferred order. If you (costly) signal tech > money > morality, but deliver money > tech > morality, you'll bleed leaderly agency.
Conversation
Facebook's existential dilemma is that it costly-signals tech > money > morality, cheap-signals morality > money > tech, but has actually failed to do truly interesting tech in the last decade in proportion with its reputation and costly signals.
Quote Tweet
Interesting rant, though I don’t agree with it. I think Facebook is a Hanlon’s razor company. Those who leave believe it is malicious. Those who stay (like me), believe it is incompetent. In a specific way — the product side isn’t strong enough to resist capture by the ads side. twitter.com/doctorow/statu…
Show this thread
3
8
Replying to
in what way does FB "costly signal tech"?
a few years ago I was interviewing at tech companies and got the impression that FB was *not* the place you go if you love the art of programming
2
1
Replying to
Big tech-focused acquisitions (you don't buy Oculus or a solar plane wifi company if you don't at least intend to be a tech company first, milk-the-ads company second). Also stuff like metaverse vision, exploring custom silicon, building an ML lab...
2
1
And "art of programming" being prioritized is not actually a costly tech signal. More "things worth high-effort programming."
1
1
Replying to
so "things worth high-effort programming" means things like "a site with high traffic that actually demands fancy tech to keep it from crashing"?
1
Replying to
That's an example, and taking commensurate bet-the-company risks. This sort of thing for eg. is not something a bank would try to do despite having equally mission critical big tech.
1
2
Other examples include problems running large data centers, all the Big Data stuff, Google's Spanner... big, high-risk infrastructure dev, where there may be little fundamental R&D to do, but a shit ton of systems design, architecture, optimization, and risk taking.
1
1
I think I’m instinctively such a hardcore “software conservative” in this sense gist.github.com/ucirello/3cca2 that I read FB’s “software liberalism” as not caring about technical quality. But the Wired article is a story about a very “liberal” technical feat.
2
1
2
Replying to
Good point. I didn't think about it from that pov. I think in some fields (eg. space travel), conservatism and quality go together and people die if you are too liberal. But software is uniquely forgiving in the range of tech risk-taking it permits.
Replying to
Yeah I’m convinced!
That is an impressive achievement on FB’s part, and it’s pure “liberalism”, and it actually worked! I was wrong; those guys clearly do care a lot about technical/engineering goals.
1
Replying to
A thing to keep in mind is that the larger the org, the less room there is for people who operate in this mode. I'd say maybe you need 4-5 such people in a company of a few thousand. The need scales sublinearly with size. So the *average* engineer will never get a shot at it
1
Replying to
Yeah these days my benchmark for a typical “cautious engineer” is a manufacturing process engineer and they’re on another planet from software
1

