3 ms·

Agreed. I think there a few key observations I've developed working in BigTech for many years now, that I don't think are necessarily obvious from the outside:

Agreed. I think there a few key observations I've developed working in BigTech for many years now, that I don't think are necessarily obvious from the outside: - You retain engineers by giving them money/stock, and by giving them a well-defined system towards advancing within the company. Typically in BigTech advancement into the senior ranks is defined largely by ability to lead major efforts that require significantly-sized teams to build out. You need a steady pipeline of junior engineers in order to keep more senior engineers on the advancement track. Orgs that stop growing have their engineers' career advancements slow/freeze and will eventually suffer attrition for it. If your org isn't growing, it's shrinking. - You retain management by giving them more reports. Power and influence within the company is largely correlated with the number of people you manage. This is both for stupid signaling reasons, but also for real practical reasons - the more resources a manager has the more they are able to take on initiatives that move their org (and themselves) forward. Everything from management to IC is aligned around continuously growing orgs. It takes a massive external shock to undo this dynamic.

rightbyte 4y ago
Sounds very unsubstainable. The amount of new staff is just insane. It has to be chaos on the inside and tens of thousands walking around without anything to do.
- redwood 4y ago
  It sounds like that would be the case but imagine every two pizza team has one or two new junior person and occasionally a senior person: you can actually absorb that at the team level in a reasonably pragmatic way. Not to say it's not inefficient or unnecessary; I just mean each team is creating its own work essentially in their narrow niche and feels it can absorb more engineering bandwidth into that niche... whether this is efficient in a macro sense is a separate question
  - rightbyte 4y ago
    Sure, but I guess many teams are "full" and there has to be newbie teams. Also FB has been hireing way more than I thought for a long time: mhio posted further down: 2017 20,658 43% yoy 2018 30,275 47% 2019 39,651 31% 2020 52,534 32% 2021 63,404 21% 2022 83,553 32% Between 2016 and 2018 a 4 person team would go 4->6->9. Assuming even distribution of new hires. Any team bigger than 4 would probably need a split. I just don't see how any organization can grow this quick without cracking big time.
    - redwood 4y ago
      Newbie teams get seeded by veterans tho
- redwood 4y ago
  It's true tho that big companies that never really let floaters go tend to have tons of them around
- potatolicious 4y ago
  Ehh, you'd be surprised! I suspect very few engineers are "walking around without anything to do". I think modeling the median tech employee as someone who wants to sit around and do nothing is inaccurate - most want to do something productive or fulfilling. Left to their own devices the average engineer in the average tech corporation will trend towards doing something. The real trick is whether or not that thing is of high value to the company. You can think about work almost as a gas - it expands to fill the container it's in. Which is to say, if you hire more people, those people will come up with work to occupy themselves. And I don't mean this in a "people are stupid and they will work on stupid menial shit" way - they will actually (in the median) try to come up with something they think is meaningful and important! The problem is that this "bottom up" approach where people at the leaf nodes of the org tree come up with their own work requires intense curation from the top, otherwise the natural tendency is for lots of teams to form to very excitedly ship a lot of stuff of dubious value to the business. tl;dr: The problem isn't engineers sitting around idle, it's that unguided engineers tend to get very busy shipping stuff that doesn't matter, or may actually be destructive to the business. Side note: Google is like this internally, and I suspect (having never worked at FB) that Meta is likely similar. Upper management provides very little product guidance, so figuring out what work should be done and what they should ship is decided very close to the leaf nodes of the company. This IMO goes a long way to explain why Google cancels so much stuff - many, many engineering hours are spent shipping products that do not matter, and eventually the company has to wind it down.
Eddy_Viscosity2 4y ago
Sounds ripe for peter principle problems. Don't these companies also allow for advancement in place?

rightbyte 4y ago

Sounds very unsubstainable. The amount of new staff is just insane. It has to be chaos on the inside and tens of thousands walking around without anything to do.

redwood 4y ago
It sounds like that would be the case but imagine every two pizza team has one or two new junior person and occasionally a senior person: you can actually absorb that at the team level in a reasonably pragmatic way. Not to say it's not inefficient or unnecessary; I just mean each team is creating its own work essentially in their narrow niche and feels it can absorb more engineering bandwidth into that niche... whether this is efficient in a macro sense is a separate question
- rightbyte 4y ago
  Sure, but I guess many teams are "full" and there has to be newbie teams. Also FB has been hireing way more than I thought for a long time: mhio posted further down: 2017 20,658 43% yoy 2018 30,275 47% 2019 39,651 31% 2020 52,534 32% 2021 63,404 21% 2022 83,553 32% Between 2016 and 2018 a 4 person team would go 4->6->9. Assuming even distribution of new hires. Any team bigger than 4 would probably need a split. I just don't see how any organization can grow this quick without cracking big time.
  - redwood 4y ago
    Newbie teams get seeded by veterans tho
redwood 4y ago
It's true tho that big companies that never really let floaters go tend to have tons of them around
potatolicious 4y ago
Ehh, you'd be surprised! I suspect very few engineers are "walking around without anything to do". I think modeling the median tech employee as someone who wants to sit around and do nothing is inaccurate - most want to do something productive or fulfilling. Left to their own devices the average engineer in the average tech corporation will trend towards doing something. The real trick is whether or not that thing is of high value to the company. You can think about work almost as a gas - it expands to fill the container it's in. Which is to say, if you hire more people, those people will come up with work to occupy themselves. And I don't mean this in a "people are stupid and they will work on stupid menial shit" way - they will actually (in the median) try to come up with something they think is meaningful and important! The problem is that this "bottom up" approach where people at the leaf nodes of the org tree come up with their own work requires intense curation from the top, otherwise the natural tendency is for lots of teams to form to very excitedly ship a lot of stuff of dubious value to the business. tl;dr: The problem isn't engineers sitting around idle, it's that unguided engineers tend to get very busy shipping stuff that doesn't matter, or may actually be destructive to the business. Side note: Google is like this internally, and I suspect (having never worked at FB) that Meta is likely similar. Upper management provides very little product guidance, so figuring out what work should be done and what they should ship is decided very close to the leaf nodes of the company. This IMO goes a long way to explain why Google cancels so much stuff - many, many engineering hours are spent shipping products that do not matter, and eventually the company has to wind it down.

Eddy_Viscosity2 4y ago

Sounds ripe for peter principle problems. Don't these companies also allow for advancement in place?