> Identify and recognize the difference between one-way doors (reversing decisions here is difficult and expensive) or two-way doors (you can make a choice and if it turns out to be wrong you can go back and pick a different door with low risk). When dealing with two-way doors it's fine and desirable to iterate quickly.
> When a decision has to be made, "disagree but commit". After a discussion takes place, even if you disagree with some aspect, you eventually commit to supporting whatever plan has been made.
> Another key take away was that previous generations need to set up the infrastructure that makes it possible for the next generation to innovate. Jeff talks about how Amazon wouldn't have been possible without existing mailing infrastructure and the Internet. He says that he's hoping that his rocket company can help set up the infrastructure necessary for humanity to continue growing and expanding in space.
Two-way doors always bothered me, because reality isn't black and white. Most decisions I've seen given fall in the grey space. Invariably deciding whether a problem was a 2 or 1 way door was politicized. Managers always argued for tech tech debt "because we could walk it back later so it's a 2-way door". Of course when such decisions backfired, it was always the engineer that "didn't Have Backbone" or "failed To Be Right A Lot", with leadership concluding that was a 1-way door and now everyone just had to "Disagree And Commit" instead of paying back the debt.
Over my 9 years there, I saw every LP used towards politics or malicious compliance. Even reading them now, it all comes across as "just so the right thing"... written in a way that sounds like there is some deeper truth.
Totally agree on the abuse of LPs in general. Abuse of "disagree and commit" is a pet peeve.
Two-way doors are often misunderstood imo. It's not about the decision it's about the cost of reversal. If you have Option A and B, then sometimes the cost of implementing A + validating A + business cost of A being wrong + reverting A + implementing B < bikeshedding the A-vs-B decision in advance.
Everything's grey. The principle is about realizing which side of the inequality it's probably on.
To combine with previous "techdebt" mentioned. I dislike tech debt, as term because it causes false analogy. Prefer friction, which is how much harder/longer development becomes as compared to your original estimate.
While observing friction you get better idea and can consider actual cost of techdebt. There are many situation where you may care less for techdebt, you have a dev script that looks bad, but you likely just gonna replace it fully when time is right.
Yet friction is painful as it slows you. Debt analogy perhaps becomes more viable now, with high interest rates. This is what tech debt is, its not debt on 0% interest rates, its debt in high interest rate scenario, everything slows down.
Amazon's "leadership" principles are related to leadership in the same way a jellyfish is related to Smuckers.
That's not saying they don't serve a purpose or don't influence the culture, just that their title is a bit Orwellian and "Be Right A Lot" is a pretty shitty principle for any culture.
I think my main criticism of Amazon would be this point:
> Some companies rely on metrics which they do not quite understand. Metrics are proxies for their in-world representations, not actual factual truth indicators.
Not just in the literal sense, though I do think Amazon labors over metrics more than any other company I know, but in the corollary sense that process a means to achieve a goal, not the goal itself.
So many times with Amazon and management that comes from there they really only know the Amazon process without understanding why any of those things were done in the first place. They're the proverbial cook cutting the end off a roast because their grandmother's pan was too small 50 years ago.
This happens at all software companies but Amazon seems to be all about process over people and generate managers who know the process even if they don't know why it is the way it is.
> "Be Right A Lot" is a pretty shitty principle for any culture.
strongly disagree.
The difference between competence and incompetence is how often you're righter than wronger (it's a spectrum of course).
mindset is one of the most difficult things to fix, but the correct mindset is going to allow you to be righter more often and that is how you rise above those who are not.
_anyone_ can be right occasionally in the same way that anyone can hit a series of keys on a piano correctly. But a master pianist is going to be able to do it much more often.
I think what you're getting at is that principles (and systems) are only as good as the individuals tasked to operate them. I too saw a lot of the same of what you mention during my time there -- people parroting the surface level jargon and really abusing the concept without ever cutting to the core of what the concept should have lead them to if they faithfully executed on it. Unfortunately, I don't know if there's a good way to solve for this as a company scales to Amazon's size.
Amazons leadership principles are great actually, I still have a card sized copy with me years later.
Thing is, as all more culture realted things, they are a two-edged sword and up for interpretation. And that allows to use them in a malicious way (this cultural aspect, PIPs and the preformance review cycles, is really bad, I agree). A ls always, just how those principles are applied depends a lot on leadership, some are great other not. Amazon's problem is, the normal bell curve of leadership talent gets tranformed into a bath tub curve by their hirong and promotion standards: You either have great people at the top end or bad ones at the other wothout any mediocre ones balancing and cushioning this.
> When the data and the anecdotes disagree, the anecdotes are usually right. Meaning that a profound analysis on data is a must.
Weird to see these notes on the front page of HN, just literally two hours after I spoke about the exact point mentioned above.
I've watched the Lex-Bezos podcast a couple of times now, and that is the point which really resonated with me.
Perhaps a simple example: API monitoring shows no issues. Isolated reports of a specific failure type, such as a gateway timeout or something that could just be a one-off. Then more of the same type of (customer-reported) failures occur within a short time-period, say within twelve hours, and all monitoring still shows "all good".
Nah, the customer is (mostly) always right, and there's very likely something bad happening somewhere in the stack. It's quite natural that the watchers would rather dismiss the whining coming from places higher up in the hierarchical chain. But when the whiner is correct, the reverence of said whiner levels up and some other telemetry gets added for better alerting improved platform robustness.
EDIT: Just to be clear, I'm not a Bezos fan, and I think anyone who is fine with having accumulated that much wealth, should perhaps think a bit harder about, things, to say the least.
> Just to be clear, I'm not a Bezos fan, and I think anyone who is fine with having accumulated that much wealth, should perhaps think a bit harder about, things, to say the least.
I think it's always useful to treat a person's general intelligence, skill in a particular field, and personal ethics as orthogonal properties.
He didn’t accumulate wealth, he built a company that successfully served customers.
It’s not a zero sum game where he took all this money from someone else. As one example, AWS has been insanely good for startups and given a lot of value to people other than Jeff and Amazon.
The one easily implementable item from this list that just never seems to catch on is: prepare a bullet-point agenda in advance, then make sure everyone reads it before the meeting, or in the first 10 minutes of the meeting.
This one is kind of mysterious to me, since everyone who has experienced it says it’s great, yet it’s still not a common practice in startups, all these years later.
It's fantastic. And if you do it digitally with some shared word processor that supports commenting, you can run a meeting very effectively.
Everyone reads, adds comments as they read and then the rest of the meeting can be settling debates around the comments now that everyone is on the same page.
As a bonus, this system lets the quieter voices speak up. Someone unassertive and shy can put a comment and it get heard
I too picked up on that line. But it kinda depends on the meeting doesn't it?
Of course an agenda is ideal. Of course each participant should come to a meeting understanding what it's for, where real, important, things are discussed, concensus reached, and decisions are made. I go to some of these kinds of meetings every year.
This year I've had the joy(?) of consulting in two corporate environments. Where the folks executing a project get to have regular (frequent) meetings with middle managers to "report on progress". What did we do? What do we plan to do? When will it be finished. 10 people are invited to each meeting. 3 give feedback. 1 "runs" it (and tries to keep it short). 6 (I'm sure) are working on other things while the meeting happens.
So yeah, no, I'm not spending 30 minutes preparing for your meeting. I'm barely attending your meeting. (And if my name is mentioned I'll likely ask you to repeat the question.)
Which brings me to my point. Calling every gathering the same thing (meeting), then prescribing advice for all meetings, is unhelpful because not all meetings have the same value to begin with.
I think bullet points are exactly the thing Bezos was arguing against. No lazy bullet points, but instead a well-written text with a cohesive narrative, consisting of full sentences and paragraphs.
I am more and more convinced that democratic decision making is necessary in companies - literally voting by employees - maybe as swiss style referendums or plain old vote for the CEO.
It's such a huge change it simply does not fit in most brains, but same is true for say social democracy during the time of Watt Tyler.
I think the point is politics exists no matter what the rhetoric of the elites says, so you may as well surface the politics keep it in sunshine and deal with it.
Plus it's ridiculous to grant a tiny elite the power to allocate resources of society simply because they got promoted...
> I am more and more convinced that democratic decision making is necessary in companies - literally voting by employees - maybe as swiss style referendums or plain old vote for the CEO.
If this lead to more successful companies, we'd see more successful democratic companies.
As things are, it will take an extraordinary amount of evidence for a claim this extraordinary.
>If this lead to more successful companies, we'd see more successful democratic companies.
If something hasn't been tried at a wide-scale I wouldn't say you can conclude that.
Concluding democratic systems don't work in the 15th century would be somewhat similar, because you're living in one of many successful kingdoms, with not many successful democracies to point at.
There were some past democracies, like the Greeks, but naysayers would say, if that system worked so well, we'd see more of it.
I think it might be worth a try. It might fail.
I personally think it's a question of scale. We know from high-level politics that usually for large nation-states, democracy works better long-term than dictatorships. Dictatorships often work well in the short-term due to faster decision making, but long-term they lack the error-correction and accountability of a democracy.
A company that executes too slowly might not be successful. So I can imagine that small companies work better as dictatorships. They need fast decisions, they need a vision.
But big corporations? They move slowly anyway. They have a lot of subdivisions. One part doesn't know what another is doing. It could be that a bit more error-correction through democratic participation helps.
> If something hasn't been tried at a wide-scale I wouldn't say you can conclude that.
I dunno; there are things that require herd adoption to succeed, but something like this doesn't need scale to show improvement.
> A company that executes too slowly might not be successful. So I can imagine that small companies work better as dictatorships. They need fast decisions, they need a vision.
> But big corporations? They move slowly anyway. They have a lot of subdivisions. One part doesn't know what another is doing.
Big corporations move slowly through decision-making because they are sampling many inputs. Yeah, things fall between the cracks because that is the nature of a large body of people, but in big companies especially, every company-wide decision is made after sampling input from almost every department (HR, Legal, Accounting, Marketing, Operations, etc).
No big company CEO decides on something like changing suppliers, changing supplied products, changing production (and produced products) without having the information from dozens of other executives, who, in their turn, are sampling information from HoD, senior management, and more.
All the executives, before board meetings, need to have their deck up to date. They never appear before the board (or the CEO) without data.
It almost looks, to me, like a federated democracy or a republic already.
Just as US citizens don't provide direct input to who the president is, they provide a representative who represents them, in big companies managers represent the people under them. The people under a manager can't vote for line-managers, but in every case that I've seen where a manager was unpopular with their subordinates, they were replaced.
> It could be that a bit more error-correction through democratic participation helps
I dunno what "errors" this error-correction would take when performed by foot-soldiers. As important as janitors are, should they have a say in determining what next years product line should look like?
I think you (or maybe me) has a view on how decisions are really made at large companies that is divergent from reality.
Yeah of course each layer has a understanding of "their" area, but the point is that if they decide to screw over their constituents then the constituents a) don't necessarily get to find out, b) cannot do anything direct about it c) face reprisals if they do
Those three points are the basics of democracy- not is my congressman faithfully representing the views i voted for.
Put in place open budgets, in-house journalism and reporting, and if the board director chooses an approach they cannot adequately explain then yes I think voting for different boss may well be an effective measure.
Will it work? It's not clear. Is the current system perfect? No? so what range of options we have, this is in that Overton window.
>>> As important as janitors are, should they have a say in determining what next years product line should look like?
I mean, apply that to Janitors having a day in who should be their next President, and let it rattle around in your brain for a bit.
We have seen so many companies happily go bankrupt, while top managers cash out big time. Few people give a shit if company is successful or not as long as everybody can make bucks until the party ends.
Democracy keeps those with power honest and in-check. Especially with technology you can enforce top-down mechanisms unlike in previous human times. Regardless of whether your policy makes sense or not.
> grant a tiny elite the power to allocate resources of society simply because they got promoted...
They can only allocate the resources of the company, not society. And they get selected based on their track record of effective allocation of those resources.
Democracy has never demonstrated efficient allocation.
Look up the "Totalitarian Bet" by Guardian columnist.
The basic idea is that in a
totalitarian society elites hoard knowledge and access to technology to support their own position, the common example is Saudi Air Force pilots are basically all princes.
democracy and openness and metritocracy leads to better people in positions of influence, and leads to organisations that are more mission focused and more flexible.
But it's a bet. There is no reason to believe that the military industrial complex of democratic mid-century USA is down to democracy or just luck.
But is "our society" defined by liberal democracy? And if so place our bets on liberal democracy
There's little indication that democracies are more flexible. The politicians spend all their time arguing with each other, eventually reaching an ineffectual compromise.
"more mission focused", I seriously doubt. More like "everyone pulling in a different direction."
> And they get selected based on their track record of effective allocation of those resources.
I've seen too many managers fail upwards to truly believe that is true in general. In plenty of companies connections are just as or more important than being effective at allocating resources.
This is what motivates me in my ... admittedly belief.
Of course politics will always play a huge deciding part in resource allocation. But if the allocation is based on a small "elite" then it looks a lot like historical societies - and we think we have moved beyond that.
If we believe in democracy for our civic society, if we believe in the Totalitarian Bet, why not apply democracy to the companies that are too big to fail?
It is also why I suspect founder led companies do very well (if the founder is good enough) - the internicine politics just gets overwhelmed by the founder picking and choosing - but history has taught us that it's vanishingly rare to find such people, to place them in the elite and have them continue to make excellent decisions over decades.
Transparency and competition drive efficient allocation, which are things encouraged in a _true_ democracy. All other forms of governance are strictly more inefficient.
That's more or less my argument in the EU - it's something federal USA has stopped noticing but it's clear in Europe.
You can become leader of Hungary by wrapping yourself in the patriotic history of the mid-European medieval struggles - struggles that carry cultural weight in the countries that bore the brunt of the middle ages. But talking about medieval knightly charges just gets a laugh in Sweden whose cultural resonance is more gunpowder and forest rights.
Any pan-European agreements need to be appealing to commonly held incentives - those almost always look like "greater common wealth, growth, consumer protection"
The Federal US has a similar setup and I think can be found at root of many of the wedge issues troubling the US
Democracy is not required for a free market to function well.
Running a company as a democracy pretty much requires every employee to be an equal owner of it. That doesn't work well. Sometimes a fearless leader, like Jobs, Gates, or Bezos, who gets things done without endless debate and compromise.
Who defines what "functions well"? And how do you define "free market"?
I'm not being pedantic, these questions require important distinctions.
Arguably, the US has the "freest market" out of all, but it also has rules and is not entirely "free".
You can make the case that you only have a "market" if you have transparency, many buyers and many suppliers. Our economy nowadays becomes less and less of a free market. Big companies buy off competition, reduce the amount of suppliers as well as transparency. They don't want an informed customer and they don't like competition. See where it got us.
Democracy is absolutely necessary to maintain a market.
Makes quite a lot of sense if you think about that. Most juniors could change their perspective cause someone in the meeting who they admire says otherwise.
>(Jeff Bezos): "Your most junior person can overrule your most senior person, if they have the data to back it up. In your meetings, the order of voice should go from the most junior person, to the most senior person."
Wholeheartedly agree!
Phrased another way:
"Data is senior to seniority."
(At least, let's presume that it is, or should be, 99 times out of 100 -- 99% of the time -- barring exceptionally rare and unusual situations and circumstances. Accumulated Wisdom over a lifetime sometimes trumps Data ("What is the exactly right thing to do given this situation, given this set of circumstances?") -- so Wisdom > Data > Corporate Seniority, although in an ideal world and under normal circumstances, all of them should be in exact agreement...)
>As a company/team, you need to seek for ways/mechanisms that enable your employees/folks to tell the truth.
In my old company, we used to have team-seeking meets every week, which consisted in a stand-up 10 mins per employee. It was quite handy cause we learned to tell the truth in benefit of everyone there.
It's about slowly demonstrating that such things are highly held principles within an organization. That is, if there's resistance to open and honest meetings within your teams, most probably it's because historically open and honest has not been a safe bet within that organization.
I've worked in many companies/teams that gave lip service to the truth and open/honest discussions, but it usually only took a few times ignoring these principles to kill any trust in the idea. Rebuilding that trust takes time and usually starts small -- it requires demonstrations that persons with power in the company will do their best to protect against repercussions for objective, good faith open reporting (the truth). This will take awhile from my experience, and in many cases there is a need to define 'good faith' a bit more, else truth sessions typically end up as emotionally charged finger pointing sessions.
It's possible to do all of this, and not even hard, it just means that those who are asked to lead will need likely need to deal with some emotionally charged arguments for a time if/when someone uses the same policy for emotional venting. Nothing wrong with a bit of venting, leaders should be able to handle this gracefully also, but my experience is a lot of leaders are not trained or prepared on how to handle such situations, and the tendency is to rely on the authority of their title rather than address the concerns directly.
Basically, there isn't much to it, it's just a rather awkward path that is best started in smaller groups, establish clear and unambiguous boundaries [0] for such a culture, and provide new reasons for people to trust in _this_ system which has likely failed them historically.
0 - I wouldn't read too much into what the boundaries should be, it's something that is dependent on the people you work with. For a team I was asked to lead, it was set early on that openness and objective discussion were essential -- if a project or situation turned sour, it was established that the important thing was to know and understand why it happened, not to punish, and that meant giving freedom to the team to make mistakes and that while we would review and correct mistakes, the team would also be protected from those "out for blood" over the mistake. It was probably 2 years of ugly arguments with other teams who wanted a scapegoat, but the end result was the team I was asked to lead grew very fast professionally and we got very good at self-correcting and owning mistakes when they happened. Boundaries for the team were that "sometimes" there are going to issues we must handle with a few "we really must avoid messing up here" clauses included -- being extremely judicious with the use of such clauses and ensuring they were truly the exception and not the standard helped retain the sense of importance for such clauses while maintaining the team's freedom to handle projects in the "best" way.
Nice. So sharing some context. I've found that it is not just about the company but org and teams within.where it can vary a lot. Even in faangs there is a very high level of lip service and worse the amount of passive aggressiveness and gas lighting that goes on is unbelievable. It's almost like systems are in place to track this and get back to you at the right time! Very mechanized.
> That is, if there's resistance to open and honest meetings within your teams, most probably it's because historically open and honest has not been a safe bet within that organization.
Perhaps to state the obvious: all these lessons mostly pertain to making large companies with history, processes, and bureaucracy/coordination costs a little less stuck in self-serving projections.
Indeed, most of the Amazon business heuristics published elsewhere pertain to rivalry within and among groups -- lessons learned from Microsoft.
Actually "building business value" means making the right decisions, and not making self-serving decisions is not good enough for that.
As a data person, I read it as: you don't always have anecdotes. So your data needs to account for blind spots and have enough granularity to see the issues.
And focus more on negative metrics / finding and addressing anomalies instead of vanity metrics and showing "up and to the right" charts to leadership.
Amazon, both globally and locally, has a horrible reputation for abusing the ever loving shit out of employees at every level. Does it build value in the process? Absolutely!
Forget politics; do we really want another company like Amazon?
Again, lessons are outside of what Amazon brings to the world. Those lessons can be applied into almost all type of business, and that's the value which bring.
Note there are no advices regarding, exploiting, abusing or not-paying extra hours.
> At 18th I've received an email from Amazon, that in order to get my package I need to "share some personal details" for a 3rd party (Aramex). No refunds in the email. I did it, naively, and no answer since that.
My guess is that the info is for the customs process between Aramex and your government. It hasn't been super-rare for me for customs procedures to sometimes take a bit more than a couple of days. It's not like the shipping company can rush the government; the government will probably not care for their deadlines.
If you know the shipping company, you may have been given the company's tracking number by Amazon, which would probably allow you to see a bit more info on the shipping progress. Here's their tracking page:
Similar experience: An order that was supposedly in an Amazon facility < 30 min from me on the 8th, was not delivered as promised by the 13th, then stated delivery between the 14-20th due to "severe bad weather" and as of today still not delivered. We haven't had snow in weeks and this is the mildest winter in several years. Now it says "if not delivered by tomorrow you can come back the next day and get a refund".
> Identify and recognize the difference between one-way doors (reversing decisions here is difficult and expensive) or two-way doors (you can make a choice and if it turns out to be wrong you can go back and pick a different door with low risk). When dealing with two-way doors it's fine and desirable to iterate quickly.
> When a decision has to be made, "disagree but commit". After a discussion takes place, even if you disagree with some aspect, you eventually commit to supporting whatever plan has been made.
> Another key take away was that previous generations need to set up the infrastructure that makes it possible for the next generation to innovate. Jeff talks about how Amazon wouldn't have been possible without existing mailing infrastructure and the Internet. He says that he's hoping that his rocket company can help set up the infrastructure necessary for humanity to continue growing and expanding in space.