More than anything, it’s surprising how well this company has maintained an illusion of competence and prestige. The general population really believes in the eliteness of their engineers. But when the rubber hits the road, the products leave much to be desired
Edit:
A hypothesis I have:
Part of the success of search is in the belief of the elite ness of their engineers. “They are the smartest so why ever try another search product”. The aura of the company is part of the brand/product. It’s brilliant marketing
Google has solid to excellent engineering depending on where you are. The problem is and always has been that the leadership hires mid HR and business talent. It’s hard to watch smart people with 40 years of experience having to jump to the tune of some MBA bro with a massive ego whose only qualification was going to the right school and a massive ego.
Google’s problem was never a lack of ability to build and run impressive engineering artifacts. It was always that the business and product side of things was run very poorly.
This just makes Google average. Execution isn't a problem at most tech companies, the problem is knowing what to build and stopping engineering teams from doing "new shiny"/resume-oriented-development.
I've been working professionally for 25+ years depending on when "professionally" starts and almost _never_ has been "the team can't build it" even in the context of companies that are basically mountains of mediocrity. The problem has _always_ been that the team didn't know what to build, and very occasionally, the best way to build it for the long term.
Execution absolutely is a problem. Being able to build something is 10% of the battle. Being able to build it well and reliably is the other 90%. A culture around building the proper tests, monitoring, and tooling to handle launches safely and plan rollouts goes a long way and most companies don't have any of that.
The difference is stark even at large companies. Oncall shifts for friends at places like Amazon and Meta are nightmares of constant pages and stress. Oncall shifts for my friends at Google have tended to be far lower key, with dramatically fewer pages and more straightforward resolutions when issues appear.
The situation speaks moreso to Google's total lack of competent product muscle. Which is a bit ironic, given that a Google PM role is purportedly quite coveted.
Apple and Microsoft appear far more intentional and successful with the products they choose to launch.
Though engineering at FAANG in general is overrated. Know plenty of underperforming people at my company that ended up there. All you needed in the past is ~average IQ and willingness to grind
I think you're insulting anyone who has worked for FAANG as someone who is not a good engineer, has an average IQ, and only got into the position via grinding. That's pretty disingenuous and not an honest assessment of the talent in FAANG and how most engineers DO want to work there.
There's a lot of smart engineers who don't work for FAANG or will never work for FAANG, but your dismissal of engineers who work for FAANG as essentially bad engineers is completely wrong and insulting.
Statistically, FAANG workers are not good engineers, mostly because Facebook and Google skews that by the sheer amount of people that they have.
SWE at Google means you're good at interviewing, and that's it. There's a reason why Go was made, because, of of Rob Pike's own admission, most new devs at Google are dumbasses. There's all the half baked products, the dogshit APIs, all working on some internal ad software that runs in O(n^n^n) because who gives a shit, we're google, we have spares in our datacenter. The same applies to Facebook, that has hired up the wazoo and has basically led to a pass-the-interview culture. Passing the interview means you're good at remembering interview questions.
I do have some praise for Netflix engineers, mostly because they remain a small-ish team with insane talent, and Apple and Amazon I can't really speak for. But just like every company, they have mediocre and average software developers.
>most engineers DO want to work there
In the past, maybe. Nowadays, you're in a pretty massive bubble if you see people still interested in joining fucking Meta. Although I guess $400k TC would make me question my morals too.
Good engineers relative to what? You're conflating averages within a company, to averages in general to software engineers in the market. And you're basing this on the failures of a product without using some baseline.
I agree that FAANG engineers aren't necessarily the geniuses they hype themselves out to be, but making broad assertions needs some baseline aside from personal experience as my personal experience with FAANG engineers has been mostly positive. That's not to say non-FAANG engineers aren't competent either, as I have worked with non-FAANG engineers as well and they were perfectly fine.
The intent of the OP was simply to dismiss anyone working at FAANG and insinuate they were all subpar and average. He thinks you can just say "from my experience" and that just absolves him from any criticism of his statements. It's completely disingenous.
I said the minimum threshold required to get hired is "~average talent and willingness to grind". This is a fact.
This statement says nothing about the average skill of a FAANG engineer. The bizarre extrapolation of statement A to fabricated statement B is quite astounding. Truly.
Not sure if it helps, but this is exactly how I read your earlier comment. I still don't quite understand the defensiveness coming from some of the responses.
If plenty of average talent people get hired at FAANG (they do, from a statistically reasonably sized sample). Then being above average talent is not a minimum requirement for joining FAANG.
The fact that is being considered a controversial statement is truly mind boggling to me.
That logic is not sound. If there's some selection criteria to select engineers from the general populace S, and a subset engineers who make it past S are average, that doesn't mea the end distribution of IQs (you're the one who brought up IQs) of FAANG engineers vs the distribution of IQs of the general SWE population would match in statistics. That is, large statistics don't just match because some subset of the populations match at some level. It's a clear misunderstanding of statistics and sloppy logic.
It's pretty absurd that you continue to think you're somehow making a logical argument and somehow think your IQ is above 120.
> "The minimum threshold to get hired for FAANG is ~average talent and the willingness to grind"
> "Some average talent people work for FAANG"
> Thus, somebody who is average can get hired by FAANG
I said absolutely nothing about the skill distribution or histogram of talent across all of FAANG engineering.
Anybody who's worked in the Bay Area has known a number of untalented people who have joined FAANG. That's reality. Believe whatever you want.
That an untalented person can join FAANG does not say anything about the talent of the FAANG engineer on average.
An interview process threshold defines the left tail of the skill distributions of the hires, and tells you nothing about the shape of the distribution beyond that. Though you can likely assume it's normally distributed.
You keep worming your way, distorting your initial statement. You initially stated the following:
>
Though engineering at FAANG in general is overrated. Know plenty of underperforming people at my company that ended up there. All you needed in the past is ~average IQ and willingness to grind
>
You have to be specific here, what do you mean "all you need"? Is this time to study algorithms? Because not everyone who studies algorithms, at the same rate, the same number of problems get an offer after an interview. Not only that, you're implying the interviews for other companies are exactly the same in regards to checking for IQ, which is completely laughable.
You simply assert that someone with average IQ that you know (and even underperformers) made it into FAANG, so you then broadly extrapolate, without evidence, that it means the selection criteria S for FAANG is simply a matter of if the person spends time studying and has an average IQ, and is no different than any under interviews.
The fact that some folks think repeating something is enough to show that it is true is truly mind boggling to me.
> If plenty of average talent people get hired at FAANG (they do, from a statistically reasonably sized sample). Then being above average talent is not a minimum requirement for joining FAANG.
Where is the study or data on that? Just repeating things and sprinkling in terms "statistically reasonably sized sample" doesn't make it true. I know you want it to be true (and it may be true) but that is not evidence for that.
While I don't entirely agree with your claim, the claim being responded to here is not what you said and opens with "Statistically, FAANG workers are not good engineers". So the reason people are replying to a statement that you didn't make is because someone else did.
Don't be pedantic, that was absolutely the intent of Arthur's post. Re-read it. He clearly insults FAAANG engineers as incompetent, then follows with FAANG engineers are just average in intelligence and grind. You're being dishonest if you think that's not meant as an insult.
Perhaps, but I choose to read his first post as saying that the startup he works for isn't capable of giving some people a successful environment, which is often true. Broadly, I see no need to prove myself, if I choose to change employers I'll be fine.
The other thing, of course, is that small companies can afford to be extremely judicious in a way that G/M can't (for example the old thing where Netflix only hired staff-equivalents). But I ultimately agree with you (like I said!) that at that point you're basically saying Google and Microsoft contain the below average engineers only from the index of current-and-former faang employees, which is sort of trivially true.
As for grind, idk I think it's good that my coworkers are below average grind. I am too. Wlb is good. Grind is unhealthy.
Smart people can get by this filter, as long as they're compliant with less studying.
Average people can get by this with more studying.
Even if you're smart, you will be studying.
"Rockstars" can't get by this filter, which is probably the intent of the interview more than finding amazing engineers.
I think it's possible that you've attached a certain weight to illustriousness of FAANG, but it's an open conversation outside of FAANG that it's not hard for an engineer of 8-10y to get a job there, the question usually becomes: "Why would you want to".
As soon as you get to a certain level in tech, your primary financial needs are met completely and you start to see the shiny veneer of FAANG wearing off.
A lot of what makes them seem competent from the outside is the sheer volume of staff, 100k people can make a LOT of shiny documents and good ideas (the borg paper has definitely made me think of working there).
The veneer falls of Facebook first "Ew, they're scooping up all that data, gross!"
Then Amazon: "They treat people like slaves and everyone I know Ex-Amazon is a burnout"
Then Google: "Everything that isn't ad-tech is a toy, sure, they're rich but you live in a massive bubble.. also "Googliness" makes it sound like a cult"
For all of these you have the implicit knowledge that you're going to be a tiny cog and overall the bureaucracy will be insane.
Anyone who has pulled back the curtain on these companies knows that while the comp is good, the engineers aren't extraordinary just because they work there or work on "x scale problems".
In fact, most Google engineers taken out of Google are fucking useless, because they've developed a body of knowledge built entirely on systems that do not exist on the outside, many of which were developed many years ago to make hard problems simpler.
Why is the defensiveness "unpleasant" when you're the one calling engineers incompetent because they work for FAANG? I quote you:
>
In fact, most Google engineers taken out of Google are fucking useless, because they've developed a body of knowledge built entirely on systems that do not exist on the outside, many of which were developed many years ago to make hard problems simpler.
>
Who is making this unpleasant? You critique a body of engineers, using biased personal anecdotes and no hard metrics, and think I am the one being negative and unpleasant? Insane.
Ok, let's break it down because if you're working for FAANG (and this is why you're defensive) then I think you're making your own case here, by not understanding what people are saying.
1) I said that hiring has less to do about intelligence than compliance (which is what the parent was saying).
Do intelligent people work there? absolutely.
Does working for FAANG mean you are intelligent? No.
This is the distinction in which you seem to be confused. The interview process does not mean that everyone working in Google is smarter than anyone else, it just means (and in no uncertain terms) that you were willing to jump through hoops.
2) I stated that there are skills you learn inside google which are not relevant outside.
Google themselves know this, it's part of why their literature is littered with "in a google context" or "a problem exclusive to google's way of working".
For example: Piper, a centralised source code storage system is not available outside Google. Everyone uses Git, or, in the games industry they use Perforce (and you will quickly learn if you work in this context that the lush SCM tools that you're used to don't exist, there might be 1 person looking after perforce for an entire studio of 1,500 people -- if you are extremely lucky).
Blaze, the build tool which was open sourced as Bazel: Does not make sense to most engineering orgs and has very low penetration in the real world.
Monarch (and its query language) are exclusive to google.
Borg, has no corollary, Kubernetes is both completely different (borrowing a few ideas) and less mature at running most kinds of workloads. You would have to re-learn this.
The statement: "knowledge of these tools is not helpful outside" is absolutely correct.
Maybe there are some ideas that if you had 5 years to redesign you could build something. Similarly though, if you took a person from today and sent them 100 years into the past and they described a smart-phone.... well, you're no closer to actually having a smart phone.
I'm simply saying that most ex-Google employees will have to learn the tools used on the outside. In some cases they're going to be annoyed at not having certain conveniences.
My experience is the same. Could be " good engineers" and "bad management" but this would not explain why the " good engineers" would work for "bad management".
If only. At least I'd only have to master Gradle's batshit insane logic. But no, between the Android Gradle Plugin which is clearly the work of a tortured soul, various tools like aapt, the Android API changes (lol MediaStore, lol version & extension checks), some tools in Android Studio which have been terrible for a while, there's a lot of bad stuff coming from Google.
But then again none of this is as painful as using Xcode so I can't really complain.
> SWE at Google means you're good at interviewing, and that's it.
Not quite true. As a hiring manager I lost several really good candidates to competing offers from Google. And these were strong engineers not only interviewing-wise.
Netflix and Alphabet engineers on par with each other, because it’s a revolving doors situation in Silicon Valley. There are under currents that move employees into different companies that have nothing to do with talent or meritocracy it’s just people hiring people they know from their past jobs.
As far as I read the comment, it is pushing back on the idea that people who work at these companies are all amazing, and does so by saying there exist under-performing people. You are at some other extreme, and so are trying to read the comment as saying "anyone who has worked for FAANG" is under-performing, which is so far off from what the comment you responded to is saying as to be a bit comical... maybe, defensive? ;P (edit: oh wow I just noticed your other flagged comments--which is why I hadn't seen them before replying here--and it is clear you are just being irrationally angry today.)
> I think you're insulting anyone who has worked for FAANG as someone who is not a good engineer
It does not matter how good an engineer you are when you produce crap. There are great engineers working for 3 letter agencies and the result of their work is people going to jail or getting killed because they are against the official narative. See Assange for example.
The fact that you interpret my statement of a required threshold to get hired as a blanket statement about the skill of all FAANG engineers is really something.
What I said was self contained and factual, nothing more.
Yes, that's all you need. But that statement doesn't imply anything about the skills of the people, just the skills required to pass the interview. Great people can pass easy processes too.
The fact that you would arbitrarily extrapolate my statement into something broader that I clearly didn't state implies to me that you're close to the bottom of that threshold.
And I'm already wealthy and pseudo-retired after writing >million lines of code for a successfully exited startup, thanks. You do you.
I used to think so regarding Microsoft, but since the WinRT reboot, deprecation of the tooling and rebooting again back to Win32/COM, while at the same time rebooting Xamarin.Forms into MAUI, and parallel to those two efforts, trying to push Blazor everywhere there is some form of Web Widget, I am no longer sure.
That's not even scratching the surface. They're spread far and wide and the ice is cracking. WPF -> UWP -> WinUI 2 -> WinUI 3 is an even bigger mess that's creating a system of frameworks that completely overlaps in functionality with MAUI, except it's Windows only - but not really.
The entire pipeline there doesn't even make any sort of coherent sense. WPF was clean, had extensive interest and support, and could have been trivially extended with more refined and "modern" styling/functionality. That's fundamentally all that UWP/WinUI are anyhow. But with each branching off, they splintered and annoyed developers in the XAML domain, to say nothing of the fact that I'm just pretending MAUI/Xamarin doesn't exist here.
I feel that the constant desire to push onto new things is, somewhat ironically, really holding back major progress. Because you put out some new awesome and really exciting tech, but things take time to mature. C# is absolutely awesome now, but was a mediocre java clone when it first emerged. It took decades of incremental and focused progress to get there.
What has pissed me most was the way they managed the C++/CX to C++/WinRT transition, because not only does WinDev expect .NET devs to keep their C++ skills up to date for specific APIs, now they want us to author components like in the ATL glory days.
Then we finally have Native AOT, but still fails short from .NET Native capabilities.
Google builds well polished products that work well. The issue is that their product strategy is bad due to the company’s culture. For example stadia as a technology and product usage wise was great. However the product placement, market growth strategy, and pricing strategy was crap.
> Google builds well polished products that work well.
The basic tech, I agree, like Youtube is great for watching videos, it barely buffers and you have billions of videos at your disposal and can watch each one and it just works.
But polished? The search is atrociously bad. "Set your location" basically doesn't work. Comments under videos are a mess, half of them are deleted, but you can still see "10 replies", only nothing will load, creators will say "I didn't delete it, Youtube did".
Shorts? Oh, the ability to upload TikTok clips and then get them presented in a way that's worse than random and actually makes you want to stop scrolling?
I don't know Stadia, but I had a similar experience with a lot of Google services. It works, the fundamentals are solid, but the UX leaves much to be desired.
A lot of these issues are quality issues but Google outsources QA and testing. Apple hires people from retail to do manual QA and testing. Automation is often not good enough since coverage is slow and triaging of results is long and tedious and very few engineers want to triage test results all day.
Are they? I mean, it's not like they have a million features and some of them haven't been tested thoroughly. Search, Comments and Shorts are central, and they suck. It's like when apple had that reception issue in their iphones, only that this is like ... "oh and you also can't read the display properly, and the speakers cut out every 30 seconds". I can't imagine them being totally ignorant and believing that their product is working great.
There's definitely testing but it's not at the level that it should be. That is, there's presubmit and postsubmit testing, but who is triaging the results? In addition, how often do we see people design tests to pass instead of design tests to fail and catch issues? I can't speak for Youtube, but some regression can be perhaps attributed to what management claims is lower productivity for new grads and mid levels, who are often assigned things like writing tests and automation (everyone in every company just wants to work on features). I think manual QA is highly underrated in general in the industry and catches so many bugs that automation just can't, and my guess is the manual QA efforts has decreased, and the discrepancy in output from seniors and juniors/mid-levels might be causing some quality issues (just my guess).
Yeah totally. They never had much manual QA. I once worked at Google and part of my job was in a rotation doing server pushes. As part of the push process I'd quickly check out the one paragraph of release notes from the releng guy which would note any new user visible features, and go and play with them on a canary server.
The frequency with which new features point blank did not work at all just blew me away. The problem 100% of the time was that some dev had written a feature given a bug ticket, written some unit tests that covered the code, but never actually brought the serving stack up and played with it themselves. They thought QA meant some test with lots of mocks, not, ya know, actually testing the feature as a user would see it. It was painful to run the servers locally so they just didn't bother a lot of the time.
What do you mean by who is triaging the results? Pre-submit failure will block your PR from landing, post-submit (e.g. integration test suite) failure will result in an incident and a likely deploy pause/rollback.
It's hit and miss, I guess. I find Google's search also worse than a few years ago, but it'll often find Youtube videos better than YT's search will. To me, YT search looks like the most naive search implementation using some synonyms and then ranking by channel popularity and upload date range, and it's getting flooded by Spam Channels (that you cannot flag and hide, because that would be an obviously good idea, and Google can't have those).
Google Docs is great for what it is, the problem is Google's insistence on building everything as a web app regardless of that decision's technical merit. Until the point where the web app architecture wets itself and gives up Docs has a reasonably good UI and the collaborative editing was/still is best in class.
That said, the speed with which the product evolves is glacial these days. OK I'm only a lite user but other than non-paged mode I'm struggling to recall what they've really added even in the last 5 years.
>Google builds well polished products that work well.
lol, I nearly choked on my spit laughing at this. The way Google Home spectacularly fails to work on my Android phone without throwing any error message, would beg to differ.
In fact, most recent Android releases have been buggy nightmares for many users pushing them to switch to iPhones in masses.
Google has some products that work very well. The others feel like abondonware in beta stages.
YouTube Kids is awful It locks up regularly, such that I have to kill the app and restart it.
They helpfully put a parental lock button right where a child would hold a phone, so they are constantly hitting the button, and bringing up a bright red screen interrupting the video.
Seems like no one really gives a shit about it, I don’t know?
I won't let my kids use YTKids anymore. Besides the app issues, people have figured out how to game their algorithm, and you end up on either an unboxing video or some gore-scare in the middle of a kids video.
My kids only get to use curated experiences where a human has actually picked the content (ie. pretty much every other Kids video app).
This seems endemic to all parental things in tech. I swear most of them seem to be done with no user acceptance testing at all. Worse, I think most of them are built by people that don't use them.
The general attitude seems to be that they can take the "build it, and they will come" approach. Who cares what people want, when you can build what you want them to want?
Had this complaint with kindle's time limits. They have it by "weekend" and "weeknight." But... as a parent, I want "school night" versus "non-school night." And those aren't the same thing at all.
After that whole debacle with YouTube and the COPPA, I sort-of wonder if YouTube Kids has to do with anything anyone actually wants, or if it's pretty much just there to serve some kind of purpose for their optics.
> Google builds well polished products that work well
No titlebar, no scrollbar, big logos using a big part of the screen, automatic updates, warning or error messages obstructing other GUI elements without a close or cancel button, etc.
Google does seem to have many solid engineers, as far as I can tell. (Though the bar must definitely be lower than back when it was a smaller company.)
Their problems lie in management: executives, line managers, product, etc.
Among the general public, maybe. From my point of view Google products are at the lower end of quality and usability. The user interface of Google products is designed by people who never saw a GUI in their life. No titlebar, highjacking right click, big logos without functions, no menus, no scrollbars, error messages obstructing GUI elements without any button to close, status mesages obstricting text input fields.
"You are in power saving mode. This function might be affectd." Is it or it is not affected ? What is the result ? On a 1.8 GHz proceesor with 4 GB of RAM this is left as an exercise for the reader.
My time playing with Android NDK showed me that they surely weren't there, I ended up wishing to have my Symbian C++ dev experience back, and many know how not straightforward it used to be like.
No, I tried them all, I always come back to Google. It guesses what I want and if not drives me towards guessing what it wants. It's the best software I use daily, it doesnt feel like an illusion. I hate that I love Google.
Area120, even with the previous cuts, was full of talented people with a background in building businesses and VC investments. It was a pretty unceremonious end to what was originally considered an important part of Google's innovation strategy.
Separately it'll probably get less coverage than things like the incubator shutting down, but I suspect more external impact will be felt as a result of them firing the head of Open Source who's been there for 19 years. :/ Tech press hasn't picked that up yet though.
Google doesn’t need an innovation strategy. They innovate lots, constantly.
Google needs a strategy strategy. They have corporate ADHD and shiny object syndrome at a massive scale.
IMO they are perpetually confused because of the fundamental conflict between a culture of brilliant creative people and the core corporate mission to make ads more pervasive in our lives.
That "look the line goes up so we must be doing good" is what caught Kodak. No one ever deserves success, and companies that stop focusing on being good will get caught slipping by more focused competitors
It is uniquely moronic that someone will get up on HN to criticize an organization that all but literally prints money. $2400 net profit per second. It's not like they are decelerating or even coasting, either: their income doubled last year, up 500% in 4 years. All those guys must be just dumb as rocks! Totally average overrated engineering org!
I don't believe that but if you do that contradicts the meme scattered around this thread that Google just can't figure out how to operate a business. Either they are inept or they are the omnipresent default choice of billions of people with monopoly power over multiple large marketplaces. But it can't be both.
Oil happens to be under UAEs ground, whereas Google has to work and compete for its ads business (vs Facebook, Amazon, others.)
Also, UAE and Venezuela both have oil but one is doing much better than the other. How you exploit your assets (oil or historical position in the ads business) matters a lot.
Wildly offtopic, but not all oil is the same. The oil in Venezuela is tar sands[1] so is expensive to extract. In fact the marginal cost of an additional barrel in Venezuela ($20) is 3x that of UAE ($7), and that's not considering the other issues involved[2]. It is then heavy oil which means it has more bituminous, long hydrocarbon chains with more sulphur so it produces a cheaper grade of product and requires more processing.
Then there is the fact that their oil company PdVSA has been under US sanctions which can't help much[3].
Despite. Most of their revenue still comes from their original product that launched 24 years ago and I think only the search engine and Youtube are profitable at all.
The ads are successful. They have a few successful vehicles for serving ads. I can't think of anything they started in the last 10 years that has turned out to be successful and it's not like they haven't tried.
Disagree. Profitability is a trailing indicator. That objective measure would say to cut all R&D.
And there are plenty of objective measures: gross margin, EBITDA, ROA, ROE, lots more. And those are just financial measures! Installed base, CLV, NPS, there are many many objective measures.
If you assume markets are roughly efficient then eliminating all R&D would cause stock prices to tank (assuming R&D is important to the long-term success of the business). I think many people on HN are biased against business concerns (as opposed to technology ones) and wrongly think they could run a business better than they really could.
Profitability does not imply that any value is being generated beyond the profit, in particular when you factor in all the externalities. Conversely, generating value also doesn’t imply that the enterprise will be profitable. The correlation is rather limited. The art is finding a way to both generate actual value and also be profitable.
Sounds an awful lot like the path Yahoo! went down. Amazing products both built and bought, a strong focus on improving the Internet ecosystem in general... then scattershot sunsetting of beloved projects, key people getting unceremonously dumped as part of broad "refocusing" efforts - where Google is now -, leading to the gradual (then rapid) decay of the entire business, changes in leadership and objectives until the company is a jumble of unconnected parts and it eventually gets sold piecemeal to less bureauctic companies.
Sundar has been hired by the voting share owners, Larry and Sergey to run their company. This is 100% their fault for being uninterested in the company they control.
> …I suspect more external impact will be felt as a result of them firing the head of Open Source who's been there for 19 years. :/ Tech press hasn't picked that up yet though.
Wow. That's a little bit surprising but not entirely. Chris' role could easily be put in a "non essential" bucket because it's an advocacy position more than anything else (unless he's doing something different these days). I would be surprised if didn't have advanced conversations though.
A120 was never considered important to google innovation strategy.
In fact what did it accomplish for the business. Google would have a better roi giving the money to Google Ventures or a16z. Or starting a private equity like firm adding notable startups to the alphabet portfolio.
> Google would have a better roi giving the money to Google Ventures or a16z.
Is there any evidence that Google is better at managing money than its investors, say Warren Buffett? If not, they should raise their dividends instead.
Google is likely not better at managing their money than investors. But once you issue dividends there is an expectation the dividend amount remains. Google did do share buybacks which is generally a more efficient way to return money back to shareholders.
Alternatively if google is viewed as a growth tech stock. I think people would trust google to make investments into growth tech startups. Google ventures performance is pretty decent.
Very little. In fact, in my interactions with them when Google Ventures were looking at a possible investment in a company I worked for I would say they stood out as by far the least technically-competent VCs we spoke to, and we spoke with everyone you might expect from the valley as well as several family offices and other investors.
At least from the outside, Area120 looked like it existed mainly so that Google could have an option for retaining key staffers who might leave to do their own startup. Instead Google could offer the chance to be a founder -- at least in terms of creating a product and managing a team.
Sundar Pichai is just a bean counter. His innovations are a botched attempt to enter China, throttling promotions in engineering, low balling non FAANG experienced devs when hiring, lay offs, lowering benefits and plastering ads on anything and everything.
I really think things started heading south when Sundar became CEO and proceeded to exchange much of the leadership.
He took over a company that was at the top of its game and fucking ran it into the ground in 4-5 years. He’ll be remembered as the worst tech CEO of a major company for years to come.
I couldn’t believe how quickly and radically Google changed and after a while, many of us cut our loses and left. This layoff got rid of many other long-timers for what seems to be no good reason.
We can only hope that all the solid engineers gravitate toward a company where the leadership is not completely incompetent.
> He took over a company that was at the top of its game and fucking ran it into the ground in 4-5 years. He’ll be remembered as the worst tech CEO of a major company for years to come.
the Google in-house incubator responsible for products such as Checks, Tables, Stack and ThreadBite
Never heard of them. Looked at them each now - nope, no bells ringing.
Separately, isn't an inhouse incubator a clear sign of a conglomerate that has gotten too big and slow? Makes me think of old telecom companies and their love of inhouse incubators, so that they can show they also do that innovative dotcom thing.
Innovation needs to happen from within, at every level.
On one hand it's basically admitting that regular development teams and managers don't have the freedom or latitude to build experimental stuff. On the other, at somewhere like an aerospace company, experimental projects are just too big and complex to do without an incubator like skunk workshop so it can be a good thing.
Software obviously doesn't inherit the physical aspects involved with aerospace experimentation (e.g. Lockheed Martin Skunk Works). The limiting factor is typically only employee focus/time.
From speaking with folks who worked in Area120, one underappreciated benefit to having an inhouse incubator in a conglomerate is that ambitious folks will go there instead of leaving for a startup.
Then they get "acquired" back into another division of the conglomerate. Startup scratch itched, equity in pocket.
So the real outcome was retention. Though that's hard to measure.
Perhaps that's why I never heard of these things - it was an inhouse holiday resort. Not meant disparagingly. I personally spent a year in such thing, before I was "recalled to duty". Corporate politics is a thing.
In-house incubators are pretty much normal these days (see ambidextrous organization). Even smaller companies like JetBrains are running them for ages. Nobody wants to miss the next big thing but you need to shield your existing business from anything that might disrupt it, so you make your incubator organizationally separate.
I wasn't there, but from what I've read calling Bell Labs an incubator seems very wrong. It was much more of a research facility funded by some mandate that e.g. 1% of the revenue of Western Electric must go to research, or something along those lines.
Yep. You can’t innovate by saying “make something innovative by copying AirTable”. At that point you’re no longer an “incubator” but a fast-follow and is a standard product development cycle.
Virtually everyone is trying to copy AirTable... and failing.
Microsoft is trying with Lists (https://lists.live.com). They released the public preview a year ago, missing key features, and failed to release final version.
Amazon tried with Honeycode (https://www.honeycode.aws/). They are no longer trying to build a spreadsheet-like interface to database.
This is reassuring. At the end of the day, google is a business. While innovation is key to their brand, they are not a nonprofit (like Apache) or an educational institution. The entire tightening of belts may be symbolic to investors across most companies, it would be jarring if the layoff were even across all businesses units.
I'd like to know which of the Area 120 projects did make it. I have been looking forward to Tables either becoming a Workspace product or getting rolled into Sheets (or AppSheet, worst case). It had just the right set of functions to replace some other tools for our team.
At this point, Google just seems to be a conventional company. The writing has been on the wall for a long while now, but their first layoff seems to cement that fact.
Even though they are not providing machine learning models, Google is still the biggest provider of high quality AI research papers in the world.
As AI models are improving, at some point they will turn into business advantages, it's just not yet big enough business for Google to make a big difference compared to ads.
"Google is still the biggest provider of high quality AI research papers in the world."
I agree that they have at least one of the top AI research groups among the corporate players. But is that going to translate into business advantage?
IBM Research was (and is still to some extent) pretty solid, but did that actually translate into business advantage? The "Watson" that competed on Jeopardy was impressive for its time. But we can look back on how the attempt to commercialize that flopped pretty hard.
Note: I'm not saying Google has IBM levels of sclerosis, at least not yet.
But there's no ironclad law that a solid research division will always turn into something profitable.
Outsiders just don't understand how ML contributes to Google's profits, because Google doesn't directly sell the ML to customers (like IBM's idiotic Watson product). But ML is plugged in everywhere: BERT (or its secret descendants) in Search, whatever unnamed inference system stands behind their Smart Reply and Smart Compose, the image segmenting and classification thing that makes searching Google Photos so easy. They even have a thing in Google Sheets that predicts what formula you are about to type. Their physical infrastructure from water pumps to cooling fans right down to CPU clock speed is controlled by machine-learned algorithms. They have a machine-learned compiler optimization pass that inserts explicit memory prefetches in key latency-critical function. ML is everywhere.
If AI is in their (production) future, what sre all those TPUs there for.. for what years now?
Everything we saw from OpenAI we've seen in one way or form from good ol alphabet already, just not accessible or obvious in their products. They're really both blessed with immense science and engineering talent and highly inert and frankly dumb business side.
Maybe, I am more impressed with FAIR and Deep Mind. Yes I know Deep Mind is part of Alphabet... that being said TensorFlow and Attention is All you Need were amazing.
We can argue whether Deep Mind is part of Google or not (I think for many years it was officially a seperated entity inside Google, and recently moved out to be under Alphabet), but basically you agree with me :)
Google is still one of the special companies that are prepared to take advantage of AI when it gets to the point of capturing trillions of dollars of value.
They're flat-footed for PR reasons more than fear of cannibilizing IMO, I'm sure they'll find a way to squeeze money out of a ChatGPT like thing. ChatGPT is rightfully received as an awesome demo, but if Google was first to release it would be evil incarnate, people will find all kinds of bad things they can make it say and it'll be a legal/PR nightmare. Not to mention the lack of "fair play" of pointing people to other sites, they already have enough flak from the little snippets, but they'd make a lot of enemies by not giving anyone else traffic anymore. Because of their size, they have a lot more constraints on them than OpenAI.
That's what Googlers think but where's the PR apocalypse for OpenAI? Everyone outside of the hyper-partisan US media bubble thinks ChatGPT is amazing and OpenAI are some kind of sci-fi operation.
Reality is Google's primary problem in doing demos of their newest models is the activism and views of their own employees, not the rest of the world.
It is cannibalizing their existing offerings. I am already using ChatGPT instead of Google for some things.
And it's not the Kodak days, Googlers fully understand that they need to disrupt themselves. And to tell you the truth many engineers there are more excited than afraid that things are happening, I think they will have the opportunity to get much more creative than before.
Were you around when Google started to get big? And all the magazines were doing cover stories either about their engineers, their vision for the future, their offices, 'don't be evil,' for awhile it seemed like a real shining city on a hill. A shakeup of the established megacorps - overnight billionaires that gave their employees naps and a percentage of their day to experiment and work on other projects.
Anyway, however obvious you mind find it, to many people watching Google become...this...is disappointing, or vindicating, or just interesting to see. I grew up with them and they're aging poorly.
Are they though. Not google's biggest fan for various reasons but most of that stuff you mentioned does still apply. The nice offices never went away and their impact on eng comp/perks has been astronomical. They didn't go evil either really, not like MS in the 90s which is what that referred to, not unless you're some sort of anti-capitalist who thinks all advertising is inherently wrong (prolly whilst collecting a nice faang paycheck). The most sinister/evil stuff they do is manipulate the public conversation around controversial topics.
Really hoping that the Tables product survives this. They were supposed to "graduate it" out of Area 120 soon, with a full v1 joining the Workspace suite, but that hasn't happened yet. We were using it extensively at my company, becuase of its native Workspace integrations. There's so much product validation already from Airtable -- why hold this back?
I don’t think huge companies should be creating brand new products. They should fund and acquire new businesses that are independently run. If you are just tweaking or improving existing products, or maybe copying a competitor’s offering, it’s possible. But trying to innovate inside of something like Google is just too hard to justify the expense. $5 million to a startup would be much better spent which is a drop in the bucket for Google employees’ salaries and benefits. Could $5 million even fund 5 people there?
I work in R&D at a big company, and I think there is some merit to this position.
It is not impossible, but incredibly difficult to innovate within a large company with business processes not designed for it.
The problem is essentially the corporate equivalent of people who post on HN saying they want to make a startup but have no good ideas. If you don't have a vision or value proposition, it is impossible to force it. Simply wanting to be an innovator is not enough.
The problem is even worse if you have management deadlines and bureaucracy. You end up simply picking the top bad idea at the next review cycle and running with it.
New businesses are constrained in the kinds of products they can make.
If you're a VC backed startup you typically have to produce results (=sales or other proof of product-market-fit) within 3-4 years, and work on a product that propell the company to a billion-dollar valuation in 15 years. Meanwhile if you bootstrap you either need very low costs or very fast time-to-market (or a very deep checkbook). And I guess if you can prove very low risk (like opening another McDonalds) you can get a bank loan, but that's a difficult route to take for product development.
Existing companies don't have these restrictions. They can finance projects that take 10 years to pay off, projects that require high volumes to be profitable (no startup could have made the iPhone at that price point), or projects that have a 3x ROI at significant investment.
In return, big companies have cultural problems that can make these things difficult. But it's valuable to have both innovation in big companies and in small flexible companies.
I tend to think the opposite - companies buying potential competition (like the way FB bought Whatsapp or Instagram) serves the public a lot less than letting these companies grow on their own.
That's sad and hope something can be done about that. But that still doesn't mean we should go back and kill the iphone if we can do it. Because then we should go back to get rid of cars (so many accidents) and kill industrial revolution as well (so much pollution)...
yes we should definitely get rid of cars, they don't add anything. you make your neighbourhood walkable and car looses to a bicycle on every metric except showing off who is rich.
Edit:
A hypothesis I have:
Part of the success of search is in the belief of the elite ness of their engineers. “They are the smartest so why ever try another search product”. The aura of the company is part of the brand/product. It’s brilliant marketing