Here is one concrete example of something that Google simply gets wrong.
They should pay external competent programmers to come in. Ask the programmer, using Google's public documentation, figure out how to do standard task X that the programmer has never done before. Say, build a Google script connector for a Looker report. Or navigate OAuth 2. Give them any necessary resources. And film it.
Do you have any idea how many ways that they would find to improve their documentation for this kind of use case? Or how much they would improve adoption of technically excellent products.
Levelset what kind of skills their customers actually have vs what their libraries and documentation assume they have. Google tends to create things that work only for Google.
I’ve always found this somewhat absurd. I get that big companies are complex and projects come and go, but it is SO bad.
I consider myself a decent SWE in the relevant domain, and after a certain amount of hours spent just trying to make some calls to Google Maps API from a non-colab python server without rolling my own interface around raw URLs, I just had to blame google a little bit. Of course, in the end there’s a beautiful dead-simple solution
these things happen (maybe not at Google) and they are called UX studies. I have been part of them and it's very enlightening for the team that builds something and understands the ins and outs to see just how poor the experience is.
This is actually one of the long term effects of these LLMs that I’m most excited about. Once “sounding smart” becomes a commodity, and anyone can whip up a 300 page smart-sounding report in 5 minutes, it’s a whole lot less valuable. Industries based on this (of which there are many) are in for a shock in the next decade and that’s a good thing.
1. They scaled up new hiring too quickly, and did so when they couldn't adequately vet applicants (around pandemic times) the company grew its headcount too much in a very short period during a windfall and did so likely without due diligence.
2. The windfall ended. The board got upset because now the company was hemorrhaging money during a downturn because they hired too many people. Leadership never had to face these poor economic conditions or pressures from a board before, and did not know how to effectively make good decisions to turn things around. So they made poor decisions. Then more poor decisions. Then more poor decisions. Their own public panic over chat-gpt then made them look worse in the eyes of investors. This led to, on top of the layoffs, high-performing, long-tenured employees suddenly feeling it might be time to retire/leave, and those that remained now have abysmal morale.
It's poor leadership in the face of change, and inability to manage the rate of wide-scale internal upheavals. It's basically just mismanagement happening at a huge scale.
At some point it felt like FAANG companies were just drowning competition by hiring everyone decent enough instead of keeping up with their high bars, but you could say that at the time everyone wanted people who could code. So much that Bootcamps started being a popular way of getting people up to speed without going through the 4yr+ education that was deemed not entirely required and relevant anymore by more and more practicing engineers in the startup world.
The not-so-evil view is that some MBAs figured each engineer was worth on average 3-5x their salary, so why not hire everyone until that almost evens out?
> So much that Bootcamps started being a popular way of getting people up to speed without going through the 4yr+ education that was deemed not entirely required and relevant anymore by more and more practicing engineers in the startup world.
Bootcamps seem unlikely to go away even without zero percent interest.
At most companies the amount of CS you need during your day job is usually laughably small. The most important question is whether you can ship, and plenty of people with CS degrees can't for one reason or another.
If you're somewhere like FAANG, you need more CS, but if you have a quantitative background it's pretty straightforward to achieve that level of competency or exceed it. e.g. anyone who works through all the exercises in Operating Systems: Three Easy Pieces[0] is likely to have better OS fundamentals than the median FAANG Staff level engineer I've worked with.
CS is pretty unusual in this way because ultimately for most of these learnings the computer can tells us whether we're right and have learned something correctly.
This is ridiculously off base and goes to great lengths to blame the workers at google instead of the management all while appealing to great-man theory.
As someone who built a billion dollar revenue division at google, I have some small authority to speak in this matter.
The issue is almost entirely that VPs’ customer is not the user, it’s sundar. And sundars customer is the board and their customer is the shareholders. The company has reached a rare position where adding a billion dollars in revenue only adds 25 billion to the market cap. But getting up on stage and laying out a very tenuous and hand wavy vision you abandon a quarter later can add 200 billion.
It’s a set of circumstances that is entirely unique to their scale of success and ultimately very toxic for actual product performance and innovation.
But I entirely reject what this writer has diagnosed here. Sorry. It’s not workers who “think too much if themselves” and I think it’s a downright shameful conclusion.
Apologies for correcting you then! I’ll let them answer, but it seems there’s a fairly obvious one: the shareholders can sell their stock for more money than they paid for it. Or do other rich people liabilities-financial-instrument wizardry to take advantage of a bigger NW on paper. But could be off base!
If nothing else, sundar specifically gets huge bonuses based on the performance of GOOG - I think his 2y bonus this year was something ridiculous like $230,000,000.
Was it $230M cash? Or $230M worth of GOOG with stipulations on when he can sell it?
Edit: it does not look like there are stipulations on when he can sell it, but he was compensated GOOG, not USD, and it was specifically formulated to be based on GOOG’s stock price performance compared to SP100:
It was also a triennial bonus, spanning some period of 2019 to 2022. Does anyone think they can come up with a better compensation formula? If so, what would it be?
I know this won’t be appreciated on here, but I figure it should be pointed out that there exists another way. My alternate compensation formula is simple: a 100% marginal tax rate above, say, $5M/y, with corresponding laws to make sure it’s not easily dodged through non-taxable compensation.
Does society really need ceos who earn many hundreds of millions? Do we care if all those people “leave the country” as the typical refrain goes? I personally say no. Let google suffer the incompetence of someone only earning a paltry 75x the average individual income in CA
I mean, I think outright saying that you don't care if extremely successful, powerful people leave the country (taking their skill, experience, and creations with them) is not a particularly strong rhetorical move. I quite like being in a wealthy developed nation with an abundance of well-paid jobs.
I’m not really trying to argue politics on here, just stating an alternate view. In my opinion, google is not successful because it has a genius ceo who is more productive than a huge team of engineers. What I can say for 100% absolute certain is that they have no creations to take with them
Thanks for the clarification, friend! I don’t think I’m an extreme left-winger (new fav term I’m gonna put that on my LinkedIn), and I don’t mind being “out of place” in a society that I think is immoral.
I’m also extremely anti-fascist, anti-monarchist, anti-authoritarian, anti-censorship, anti-discrimination, etc etc etc, and living in a country that embraced one or more of those things wouldn’t change my mind just so I could fit in. I imagine you’re much the same :)
Mostly referring to absolute/mixed monarchies like Saudi Arabia, UAE, Morocco, and of course the very worst of them all, the Vatican.
Not a coronation fan myself but glad you enjoyed! I did like the pic where he was sitting in a chair (…throne?) holding two very goofy looking scepters, tho admittedly there was an element of ironic enjoyment there lol
I don't know this specific example but I've been in many C-level discussions where similar things happen. There is rarely a belief that it will be temporary. There is usually a lot of confidence (even if misguided) that there will be sufficient follow through to quickly justify the addition.
In my experience people over-attribute to malice or cynicism what is usually just crossing the very thin line between entrepreneurial confidence and self-delusion.
The article does blame the management though, with respect to its concern over cancelled products.
> some talented executive wanted to do them, and other Googlers joined him or her.
I think you're right in there are issues in the article. I think the article is generally just trying to clarify a more abstract point: current high tech top-down corporate culture favors appearances more than the fundamentals of what makes good things good.
Thanks for taking the time to share the unique perspective.
As an ex-googler with SO much less authority: this article reads like someone working backwards from first principles, and in this case pretty boring/typical ones like “great man theory”, “the liberal elite are weak/dumb”, and “no one wants to work anymore”
> And sundars customer is the board and their customer is the shareholders
This is by design. As CEO you are supposed to have your shareholder's interests as your main priority.
And as you say, your shareholders are often served better by making upbeat announcements than making actual products for customers.
Large corporations have different dynamics than small businesses, and different goals. It's not even vaguely surprising that Alphabet has a different culture than Google did 20 years ago.
What I find surprising is that everyone seems to ignore this and think that Googlers must be awesome at their job because they work at Google. Those days are long gone.
Are you saying that google’s engineers aren’t above average in the industry? I certainly know they don’t have a monopoly on rock stars, but in aggregate do you think they have the same hiring bar as a boring f500 company or even Microsoft or oracle?
Quality of code? Speed of development? CS knowledge? Ability to interpret arcane customer requirements correctly?
What does Google's recruitment process optimise for? The ability to whiteboard an algorithm and complete difficult HackerRank puzzles from what I've heard. Is this a good set of criteria for recruiting "the best"?
Once in Google, are they learning more than their peers outside Google? Or are they learning how to survive the internal politics? Which helps them not get made redundant more?
I would expect "the best engineers" to be in small tech companies where they can make a huge difference and really contribute to the company and see the things they build get used by customers every day. Not in a vast organisation drowning in bureaucracy and politics.
As the article says, Googlers tend to think of themselves as special without any real justification.
My experience says otherwise because of Googles unique mix of
1) giving engineers experience at scale with sophisticated systems
2) solid mentorship opportunities
3) training them with the social skills that show them their code matters as much as their ability to work with their team and execute
4) resources to execute and learn about how to execute fast
5) the reality of pokitics
In general, when I’m hiring for a startup, I see snr+ google engineers as great hires. I find that other startups CEOs that try to complain are actually really saying, honestly, they are less talented than their prospective google alumn and they don’t understand the value.
Google’s interview process puts immense emphasis on doing thorough system design.
Additionally, in contrast with companies who do ask hacker rank questions, Google only asks novel, unique questions that are not on leetcode or hackkerrank, prizing creativity in problem solving over memorizing.
Finally, since everything is reviewed by a hiring committee, there’s almost no chance of nepotism or someone playing favorites for whatever reason.
When I interviewed at Microsoft, I was asked to implement numerous classic CS algorithms, not disguised even remotely. With google, none of the coding questions were anywhere to be found on the internet.
Big organizations tend to follow regression of the mean, but at FAANG i’ve yet to see see the absolute bottom of the barrel that
have often snuck in at the various small companies i’ve worked for.
Certainly small companies have top tier talent as well as google and other FAANGs, but there’s remarkably little bureaucracy at FAANGs as well. It sounds like you’ve developed your opinion from a vocal minority that have left in anger rather than first hand experience.
So the article that the article mentions [0] has the bureaucracy as a problem - risk minimisation is the main organisational criteria, so any change is overwhelmed with bureaucratic processes to ensure it doesn't cause risk (my paraphrase). Is this true in your experience? How much bureaucracy is "remarkable little"?
The changes they are talking about are consumer visible changes to the large money makers. I'm in infrastructure so I don't have first hand knowledge of the former, but that tracks from what I've heard. Making large changes to search or ads sounds like a bureaucratic nightmare.
However, making changes to infrastructure, to technical products, to workspaces, to cloud products, and to APIs seem to strike a nice balance between quality and velocity. You need code reviews, there are tests and static analysis and all that jazz but otherwise ship it. There are very few MBAs around to stall things or create unnecessary process.
the way I read it is that management is to blame. there is absolutely no way the workers - no matter how exceptional they think they are - can thrive in an environment that does not enable this kind of behavior.
I didn’t see an attack on rank and file away at all. This article was squarely aimed at Google’s later leaders ie Sundar.
He’s a peacetime general and he missed the coming war. He also has continued, like past leaders, a strategy of letting inmates run the asylum (“talent” driving their own projects without a firm connection to overall strategy and customers).
“Talent” will do what you tell it to do. My current company also believes in hiring talented people, but our CEO is also extremely clear about overall strategy and customer orientation, so I rarely find that people wander too far off in unproductive directions.
That talent you are insulting literally created the breakthrough LLM technology which then was ignored by Sundar for years.
What this author has completely wrong was that the talent actually came up with tons of great products that with focus could grow and add revenue. But that was nothing compared to vaporware on stage at IO for execs.
I think the response here shows both your lack of seriousness and experience here. And a note of elitest bitterness.
I was an snr leader, not the talent. I am freely admitting where the value came from. Brilliant workers smarter than me.
I really question why you’d argue with that when it was ICs who did breakthrough work like transformers that are now fundamentally changing computing. What an arrogant and foolish perspective.
Take a deep breath, you’re reading into it far much more than is actually there.
Nobody said anything about lack of pockets of cutting edge research. This is about products.
LLMs are a perfect example of them being squandered because nobody converted them into a useful product that could be “sold as a vision” by execs.
The people working on them were arrogant enough to think that LLMs would sell themselves and building a tool for users was someone else’s job. The product managers weren’t innovative enough to come up with a product to let people interact with LLMs in a useful way. The execs were too blind to not see the vision of AI doing everything, which is a narrative that drove nvidia to a trillion dollar company.
The people who implemented LLMs will be famous… as academics. The machine they are a part of completely failed to execute on them.
You’re asking for evidence of a lack of evidence. It would have leaked out if there was some huge vision of AI assistants that was being ignored by execs.
I am not "asking for evidence of a lack of evidence". I am asking for evidence that the engineers held this arrogant position. There is no "lack of evidence" described in this supposed arrogance.
Yes you are. No google product emerged because there was no product vision from the researchers, the execs, the engineers, and any of the PMs who read about it. It would have leaked out if the researchers had some grand product that morons at the top squashed.
Google is a massive company and cannot keep a lid on anything like this. There is no evidence that there was anything there as shown by both the lack of rumored products and the lackluster also-ran release of Bard.
This type of arrogance is extremely common in any research positions (both academic and industry) because people are trained very early on that it’s not their job to suggest how these things might be useful. Papers are not reviewed for applicability, grants are not based on it, dissertations are not graded on it, and on and on.
It’s systematic arrogance trained into researchers by the institutions.
Look at Alan Turing and many of the other big names in CS. The same arrogance is in all of their writings, speeches, etc but that’s expected because it’s academia. A researcher’s role in an R&D department at a corporation is not as disconnected as academia, and the researchers should know that.
Some organizations have the opposite problem. The culture has a baked in assumption of things going sideways and taking forever. Under such circumstances there is a learned helplessness and unwillingness to even try that is soul crushing. If you get hired in as a fresh graduate or from a less dysfunctional place over time you may come to believe that you are as slow, dopey, and unmotivated as everyone else.
Its funny, this person writes a poorly proofread article (while actually expecting people to pay for it) and then puts nuggets with completely unsubstantiated judgement but still wants to weasel out some image of being logical and objective about it. The Berkeley professors are “clearly leftist and should be ignored”, except for the one cherry picked nugget that supports your argument for the day? What’s the point of that statement? Who’s the political whack job, the person making a point with data or the person trying their level best to disregard the data because it’s from a leftist?
What’s with arguments like this becoming popular anyway? Should we be worried about it?
I loved that part! “He says he has done studies proving that rich people are mean, which is clearly wrong because he’s biased. But this is proof that people from different cultural/socioeconomic backgrounds I don’t like are broken!”
Article criticizes an general named Fredenall, saying his lack of combat experience doomed him. He was basically a military MBA type.
> After service in the Philippines and other overseas and stateside assignments, Fredendall shipped out to the Western Front with the 28th Infantry Regiment in August 1917, four months after the American entry into World War I. He held a succession of instructor assignments in the army's schools in France, and commanded one of its training centers. He built a record as an excellent teacher, trainer, and administrator, and ended the war as a temporary lieutenant colonel.
This article says look at all these great wartime generals, they had in the trenches experience, and lists Eisenhower, saying this:
> Eisenhower was desperate to get into combat, but he finally made it to France only as the war was ending.
Eisenhower did not see combat, he trained troops, just like Fredenall. He was basically also a military MBA type.
Very similar backgrounds. So maybe the difference between them was... talent? It undermines the authors point rather than making it. Most of the American generals all went to West Point! They might as well be Ivy Leaguers!
The part about the 5 extremely successful tech execs not graduating it disingenuous. Look at the schools they did attend, even if they didn't graduate. They had connections. They had family money. Gates and Zuckerberg for sure. Ted Turner went to private schools and then Brown. Jobs, sure, one in a million. Maybe Ellison too. So that was also talent? Kind of a mess here.
Basically this article is terrible. You need talent, plus agency, funding, C-level support. Seems more like they're not actually letting talent do what it's supposed to do?
Ellison didn't graduate. Nor Gates or Zuck. We're talking about top-school MBA's here, which was the topic (Enron).
As for Eisenhower & Fredendall: you can't really be serious. Fredendall was not a wartime general, nor was he a good planner. Plus, he was a coward who refused to listen to people, and disposed his troops ignorantly. This section was a discussion of wartime vs. peacetime generals, not MBA's.
I don't think you understood the article or what I wrote.
I understand that Ellison, Gates, and Zuckerberg didn't graduate, then I said they still attended top schools. They had connections. MBAs at top schools are only really worth it if you're under 25, and even then, only for the connections. The entire list has almost nothing to do with having talent or not, or having an MBA or not. The key is almost exclusively to be born into an at least somewhat wealthy family so that you can buy connections. Or if you're not rich you have to work super hard and be lucky. Very few people really believe getting an MBA requires talent. I don't really understand how this connects to most of those people being talented or not. They all became the CEOs of enormously successful companies. Maybe this list would make more sense if you said, look at these awful CEOs, they had no talent even though they had MBAs?
Neither Eisenhower nor Fredenall were wartime generals before WW2. Fredenall actually became a general before Eisenhower. Both performaned administrative work and training between the wars; neither directed troops in combat. Fredenall actually did command troops in combat, although as noted it was disastrous. Important to note, Eisenhower actually selected Fredenall to command those troops. I don't really see how Eisenhower was less of a peacetime general than than Fredenall. It's a really strange comparison to make.
The article has a bunch of points it tries to make and I just picked two of them that didn't make any sense.
> I don't think you understood the article or what I wrote.
Funnily enough you appear to be responding to the article's author. If you notice his other comments in this thread its apparent his reading comprehension is not at a level that produces productive conversation.
That's funny, because I don't think so understand what I wrote, either. Your comments
> I don't really understand how this connects to most of those people being talented or not.
demonstrates nothing but your own agenda. The article doesn't say they're talented or that they're not. It says that most of them got there without an MBA from a top school.
The part about elite schools and MBAs doesn't come until the fourth section, yet you've imported that into the rest. And family connections don't come into it at all.
Secondly, Ellison had absolutely no family connections or inherited wealth. Read his biography.
I can also tell that you don't spend any time reading military history, because the comparison of Fredendall (btw, learn to spell his name right) and Eisenhower is something absolutely no one knowledgeable would ever make. And yes, Ike made a mistake, but he corrected it.
Google has never been a product company. The last and maybe only major product success they have had after search is gmail. Everything else they bought and sold. So I don't know why people are getting shocked now that when they suddenly have to innovate, they are falling behind. Google is great at algorithms and a huge amount of data processing etc but how to turn it into a product is extremely unclear to them.
Google Maps first started as a C++ program designed by two Danish brothers, Lars and Jens Eilstrup Rasmussen, and Noel Gordon and Stephen Ma, at Sydney-based Where 2 Technologies. It was first designed to be separately downloaded by users, but the company later pitched the idea for a purely Web-based product to Google management, changing the method of distribution.
>how to turn it into a product is extremely unclear to them.
What according you takes something to turn an algorithm into a product? Google did try a lot of things, a lot of products but it didn't catch on so they had to shut them down.
> ..a professor at UC Berkeley, gave a TedX talk .. The extreme leftist bias of virtually every social science and humanities faculty at Berkeley is more than enough reason to discount this heavily. .. Nonetheless, we have to admit that maybe he’s onto something there.
This amounts to saying: "normally I'd dismiss this research based on an ad-hominem, but in this case I agree with its conclusions so I won't". Which is certainly one way to write a think-piece, I guess?
I agree with pretty much all the other comments here - I always find it amusing when an article bemoans the lack of humility in others with a style that is supremely self-assured and cocksure.
> The extreme leftist bias of virtually every social science and humanities faculty at Berkeley is more than enough reason to discount this heavily. You won’t be surprised that he found what he wanted to find
Granted, motivated reasoning may be alive and well in the academy; but this statement asserts personal motives where the author almost certainly cannot make this claim with any certainty.
They should pay external competent programmers to come in. Ask the programmer, using Google's public documentation, figure out how to do standard task X that the programmer has never done before. Say, build a Google script connector for a Looker report. Or navigate OAuth 2. Give them any necessary resources. And film it.
Do you have any idea how many ways that they would find to improve their documentation for this kind of use case? Or how much they would improve adoption of technically excellent products.
reply