These aren't really "tough questions". Still, Bezos gives a good interview and makes Amazon sound like a good place to work. He's right-on with this:
What really matters is, companies that don’t continue to experiment, companies that don’t embrace failure, they eventually get in a desperate position where the only thing they can do is a Hail Mary bet at the very end of their corporate existence.
I've seen that first hand. It's ugly, especially because the best people leave companies where they don't get to innovate and own new things, so you're doing that "Hail Mary pass" with a lot of deadwood in your ranks.
That said, I'm having a hard time reconciling his pro-innovation stance with Amazon's use of stack-ranking, which causes divisions to ossify and halts innovation. It seems hard, after a point, to innovate if you don't trust your own people. You'll have a core in-crowd that will toss you ideas for a while, but eventually they run out and lose the ability to keep pace.
That's the question that I would want to see asked: how do you reconcile your ideals (which are admirable) with the reality of a company known to use stack ranking?
(posting under new account because I don't think anyone can talk freely about their employer under a real name)
Stack ranking is not a well defined term. It can mean any of
1. The result of the review process is that people in a team are ranked, the top X% are rewarded and bottom Y% are punished.
2. Reviewers are asked to rank people they review (possibly not all in the same team) and this information is part of the input to the review process.
3. When making decisions, all the people in consideration (possibly not in the same team) are ranked.
Case 1. is a problem because we can't expect the law of large numbers to hold within a small team, and because it gives people bad incentives.
Case 2. is less of a problem for incentives because people being compared can be from different teams. This also provides comparisons across teams that let people be judged relative to the company rather than their teams, and so the law of large numbers does hold.
Case 3. is not really relevant to anything, since when deciding the outcomes for N people, you obviously want the best outcomes to go to the best people. Stack ranking is just an aide to turning subjective impressions into numbers. I've seen it used in essay grading, and also assigning final grades in classes (ranks are based on individual test scores and not altered, then cutoffs for letter grades are determined).
Stack ranking may, in a sense, exist at Amazon. But as a developer here, it's invisible to me. I write my reviews of some of my colleagues (I choose which), they write reviews of me. All reviews of me go directly to my manager, who will compile them together with his review of me and my review of myself to build a full year-end review of my performance.
Those yearly reviews will factor somehow into my pay, bonuses, etc, but I only see the final result of it. Was I stack ranked?
Stack ranking is very present and very real at Amazon, but if you have a politically capable manager you will find yourself (as non-manager) able to ignore it.
That doesn't mean that you or your manager are necessarily doing a great job though, or that other people aren't being hurt by the system.
It totally depends on your manager. My first manager encouraged me to do good work and make sure I got customer feedback for it. My new manager starts every one-on-one meeting with phrases like "not everybody gets promoted" or "the reality is that you will be compared to your peers"...
A manager I had at Amazon met his new team/underlings for the first time, and decided that a good way to start things off would be to explain in detail the process by which people are fired. I suppose he was trying to "motivate" them?
This same manager also later became notorious for angrily declaring that "<productName> has no technical debt!" in a meeting, despite the product being unable to meet any reasonable SLA for years (bad architecture, bad code quality, >100% annual turnover). Not sure I'd use him as an example of management there, but pockets like this certainly exist.
Yes. That is exactly what us happened. I know a few people who were edged out like that.
Make no mistake, Amazon is not a great place to work. Ok place to work maybe. Good if you have managed to fit in. But even after that, they underpay. Friends with literally a decade of time out in are making relatively low considering their wide areas of responsibility.
Comparing with my friends over at Microsoft, while the base salary is certainly lower, cash bonus and stock pretty much equalizes compensation, at least as an SDE I. Total compensation is lower than Google/Facebook, but that's a high bar for ANY company.
I think this myth really continues both because people tend to only pay attention to base salary and don't consider cost of living/state taxes, and due to the fact that Amazon does underpay warehouse workers, as evidenced by the current strike in Germany.
I can't speak for higher level positions though. But Amazon tends to strongly favor giving out more stock over more cash, which goes back to the same issue of not comparing total compensation.
Recruiters push the "total compensation" idea because it's cheaper than offering real money, but in my experience salary is all that matters.
Stock always comes with a vesting schedule and a forfeit clause, which means that if you accept more stock as a substitute for a better salary, you are implicitly agreeing to let them retroactively underpay you for the last couple years of your tenure, whenever you move on to the next thing.
Don't kid yourself that you're going to stay at the same company long enough that it won't matter, either; the raise you get by switching jobs always dwarfs the raise you get by sticking around and jumping through the perf hoops and trying to get a promotion. So the only way to make the stock pay out is to screw yourself on salary over the long term.
What's more, keeping a large portion of your net worth in your employer's stock is a terrible diversification strategy, so you will probably end up flipping it and investing in something else as soon as it vests. So.... what was gained by accepting compensation in stock, again? I'd rather they just pay me cash money and let me invest it however I want.
Senior people at Amazon that I personally known and worked with have said they were undercompensated compared to offers from other, similarly-sized companies. Most of the good ones no longer work at Amazon.
Also, many companies make the mistake of making performance reviews part of the transfer packet. It seems like it just makes sense (you have "data"; why not use it?) but it actually fucks up the culture. It's basically never a good idea to have past performance reviews come into play in a transfer decision.
First of all, it makes individual workers have their eyes on the scoreboard (reviews) rather than the game, because there's no escaping mediocre reviews; if you get smacked, you're stuck on a team forever. (And, sometimes you have good workers who get shitty reviews to prevent flight.) Second, it slows internal mobility to a crawl, because people getting lousy reviews can't move and people getting good reviews don't want to move, lest they risk getting mediocre scores and showing "negative trend" (making promotion nearly impossible). So, collaboration stops (due to the impossibility of lateral or diagnoal movement) and departmental position becomes permanent and tribal, and you get the "warring departments" dynamic. Even if you don't agree that departmental rivalry is a bad thing, what it tends to mean is that, due to the unpredictable sabotage events, whole sectors of the company become unreliable and upper management finds itself duplicating effort and running "bake-offs", but talented people avoid bake-off projects like the plague, so you get a "Dead Sea Effect".
Usually, when people talk about stack-ranking, they're also talking about Enron-style performance reviews (that is, performance reviews that are part of the transfer packet). You can theoretically have one and not the other, but they tend to go together.
If he were really honest, I think he would say that the innovative ideas come from the trusted core in-crowd, while everyone else is interchangeable. Amazon has a tough culture that chews people up and spits them out. I frequently hear from Amazon employees here in Seattle that their average tenure is just one year. The high turnover is not a bug, it's a feature.
I think he would say that the innovative ideas come from the trusted core in-crowd, while everyone else is interchangeable. Amazon has a tough culture that chews people up and spits them out. I frequently hear from Amazon employees here in Seattle that their average tenure is just one year.
So, the question is: how does one get into the "trusted core in-crowd"? Is this something that one can pre-select in the hiring process? I wouldn't work for Amazon unless I were fast-tracked into that set and I imagine that the odds are very long.
The high turnover is not a bug, it's a feature.
We need a third case between "bug" and "feature" for "that which works as designed, but does a harmful thing". In the short term, the "Real <X>" phenomenon (i.e. "Real Amazoners" get to work on the cool stuff and innovate, and the peasants do the shit work and burn out after a year) motivates people. In the long term, it's bad on both sides of the divide. The losers hate it (that's obvious) and begin to underperform and sabotage the organization, to the point where you get an unreliable "Swiss cheese" organization. However, it also leads the winners toward complacency and entitlement, as well as obsession over relative status, which leaves the company ill-equipped to compete on absolute terms and keep relevant. These degeneracies can be slowed down to the point of happening over 15 years instead of 6 months, but they are inevitable.
So, the question is: how does one get into the "trusted core in-crowd"?
At most companies it's based on playing the political game well and building alliances with the right people. I've never worked at Amazon but I doubt they are different in this regard. It's just how humans work in the general case.
We need a third case between "bug" and "feature" for "that which works as designed, but does a harmful thing".
The balance of what I've read does not make it out to be that great a place to work, but who knows. I'm not interested in working at large companies in general these days.
In any event, I think it's a very interesting company and one I wish I knew more about.
What really matters is, companies that don’t continue to experiment, companies that don’t embrace failure, they eventually get in a desperate position where the only thing they can do is a Hail Mary bet at the very end of their corporate existence.
I've seen that first hand. It's ugly, especially because the best people leave companies where they don't get to innovate and own new things, so you're doing that "Hail Mary pass" with a lot of deadwood in your ranks.
That said, I'm having a hard time reconciling his pro-innovation stance with Amazon's use of stack-ranking, which causes divisions to ossify and halts innovation. It seems hard, after a point, to innovate if you don't trust your own people. You'll have a core in-crowd that will toss you ideas for a while, but eventually they run out and lose the ability to keep pace.
That's the question that I would want to see asked: how do you reconcile your ideals (which are admirable) with the reality of a company known to use stack ranking?