So, here's the hidden secret to Google: every inexplicably stupid move they've ever done can be described in terms of promo packets.
Every duplicate messaging app? That's someone's promo packet item.
Every ground-up incompatible API rewrite? Also a promo packet.
Google produces new work purely to satisfy itself. Their hierarchy forms its own internal economy where promotions are purchased with headline-grabbing actions that fool managers into thinking they provide business value. We'll call this "Googlestan".
Yes, this occasionally causes problems in the external, "real" economy. Writing message apps as disposable products means you don't have an answer to iMessage or Whatsapp. Breaking APIs every 3 months makes Google Cloud a nonstarter for anything other than easily-migrated guest OSes with a more sensible deprecation policy.
But that's how Google was built, and how Google will continue to be built, because all hierarchies have a rule zero: self-preserve. You cannot build a new Google without disenfranchising the people who currently know how to game the current Google, and those people will instinctively fight against an engineering culture they do not understand.
The only exceptions to this are the core economic vehicles between Googlestan and the outside world: Search, Chrome, and Android. Note how each one of these products have a dramatically more conservative roadmap, with a reasonably minimized number of breaking changes. Hell, Chrome specifically calls breaking changes "interventions", because they're that serious about not making them. These products form a moat around Googlestan that protects the country from invaders, so they themselves are isolated from the kinds of people who would gank them for the sake of a promo packet.
[0] If you're wondering, "how does Google internally handle breaking changes everywhere without boiling the ocean"... the answer is that they have an automated ocean-boiling machine that lets them rewrite the entire Google code corpus whenever an API breaks.
<super tiny>I must be stupid, but... what's a promo packet? I never worked for any of the FAANGs (other than once, at Lab126 and accidentally inventing the worst thing ever) so I don't know all the lingo</>
Even though I'm simply trying to guess from context without actually knowing, everything you just said sounds both accurate and hilariously well put.
Edit: WOW. I had no idea that's how things (even used to) work within those companies. A packet of materials you submit to get promoted. I'm almost bowled over with laughter.
That explains everything.
Here in the normal world, I get promoted by... being good at my job. Asking for it as part of a performance review, typically one I negotiate for as part of my employment, also helps.
Promo packets were an attempt at a way to figure out how to promote people when you had thousands of good engineers and they couldn't figure out how to decide who had most earned it. You didn't trust their manager, they had a limited resource of 'money' to give. Maybe it was a bureaucratic approach, a little like the army?
I think this is an inevitable thing that happens when you have huge orgs with a limit on who can be promoted, there's not enough reward for everyone. I don't know how to do it. Trust managers doesn't scale, people reward their friends or whatever. Microsoft faced this too, but they didn't have a packet approach, it seemed like the senior managers decided, without having a paper trail like that.
To be clear, it's just you making a list of things you accomplished, because otherwise your boss is going to have to dig through their email to figure it out, and they may miss something / get something wrong which will hurt you. The lingo might be specific to FAANG but the practice isn't, including the part where you do "unnecessary" work to try to pad the packet.
> at Lab126 and accidentally inventing the worst thing ever)
Lab126 created some of the first e-ink technology, right? Is the Kindle, or e-ink screens, or e-ink patents, the worst thing ever... or is there something else I should know about in this space?
> Lab126 created some of the first e-ink technology, right? Is the Kindle, or e-ink screens, or e-ink patents, the worst thing ever... or is there something else I should know about in this space?
There's a lot to unpack in that sentence. Care to elaborate?
In my defense, I said it was a horrible idea at the time, we'd have to stream everyone's audio to the cloud to get the keyword spotting to work...
Also, Lab126 did not create e-ink! E-ink came out of the Media Lab long before the first kindle.
I'm glad that they're finally making one with a stylus, when I was there (more then a decade ago) there was a prototype tablet you could write on that had a brilliant new sort of user interface.
"collection of material you submit to support your case for promotion", it's actually not even a thing anymore which adds another layer of irony here, and he is dead wrong (see my other reply).
It's easy to bamboozle yourself from the obvious "maybe people are incentivized to do things to get promoted and perhaps even unnecessary things" to wild unrelated fantasies of how this could explain decisions you don't agree with
> The only exceptions to this are the core economic vehicles between Googlestan and the outside world: Search, Chrome, and Android. Note how each one of these products have a dramatically more conservative roadmap, with a reasonably minimized number of breaking changes.
I'd add Google Maps and Google Mail onto this list.
This isn't true and I don't even particularly care if you think it is and you're a fellow Googler. I know for a fact it isn't.
There's some trivial truth to it, of course, but specific assertions are laughably false and more complicated than you are claiming.
I very much would like to reiterate this sort of thing is unhealthy, the point I was making was people moralize while assigning grand motives to a large # of uncoordinated actions about decisions that are obviously more complicated in real life if you were making them. This sort of is a perfect exercise in that
Then can you explain to us why does google do self-owns such as 10 messenger apps?
Promo-driven development and a bias towards greenfield is something that many engineers in other large tech companies are familiar with, and articles such as this are written by former googlers frustrated with the promo system: https://mtlynch.io/why-i-quit-google/ . Put two and two together and it seems like a likely explanation in light of no additional information and a meme is born.
I come from a company that has a google derived promo system, and I believe it when promo driven development can explain a lot.
Especially when that system values certain things like new 'innovations' vs important maintenance. Or tech leadership of multi-team projects which leads to forcing migrations on the rest of the company to get multi-team points vs. a seamless backwards compatible one done behind the scenes not being a multi-team project, so you don't get promoted for doing it. Or valuing mindless metric number go up over a more thoughtful review of the real impact which, gasp, might not involve some numbers sometimes.
These systems also change very slowly and are hard to change overall. I think google still does 5 leetcode interviews back to back, right? Despite them being shown they're not very good indicators of real job performance?
I had a google interviews loop, 2 out of 3 weren't leetcode related questions. the 3rd was a easy medium bfs. one of questions drilled into teSt driven development and etc.
Every duplicate messaging app? That's someone's promo packet item.
Every ground-up incompatible API rewrite? Also a promo packet.
Google produces new work purely to satisfy itself. Their hierarchy forms its own internal economy where promotions are purchased with headline-grabbing actions that fool managers into thinking they provide business value. We'll call this "Googlestan".
Yes, this occasionally causes problems in the external, "real" economy. Writing message apps as disposable products means you don't have an answer to iMessage or Whatsapp. Breaking APIs every 3 months makes Google Cloud a nonstarter for anything other than easily-migrated guest OSes with a more sensible deprecation policy.
But that's how Google was built, and how Google will continue to be built, because all hierarchies have a rule zero: self-preserve. You cannot build a new Google without disenfranchising the people who currently know how to game the current Google, and those people will instinctively fight against an engineering culture they do not understand.
The only exceptions to this are the core economic vehicles between Googlestan and the outside world: Search, Chrome, and Android. Note how each one of these products have a dramatically more conservative roadmap, with a reasonably minimized number of breaking changes. Hell, Chrome specifically calls breaking changes "interventions", because they're that serious about not making them. These products form a moat around Googlestan that protects the country from invaders, so they themselves are isolated from the kinds of people who would gank them for the sake of a promo packet.
[0] If you're wondering, "how does Google internally handle breaking changes everywhere without boiling the ocean"... the answer is that they have an automated ocean-boiling machine that lets them rewrite the entire Google code corpus whenever an API breaks.