Considering that in Meta I heard (from friends) it is/was easier to work remotely and even easier to get into (they are less leetcode oriented) I would expect the problem with Google loosing is somewhere else. Google was loosing on many fronts for a while:
- dart lost to typescript on web
- angular lost to react
- tensorflow looks like currently loosing to pytorch - seems like google got bored and more development is for JAX, Keras wrapper [0]
- IMHO flutter will loose with react-native or kotlin compose multiplatform - compare github insights for details
Meta on the other hand kickstarted open source Llama community. In this situation it's hard to bet on Gemmini or Gemma as 3rd party developer considering google projects kill records. The only project they were really to bet on and invest for the long run without getting tired early on was Chrome and Android.
i believe this trend is a result of management culture and performance metrics that attempt to measure "impact", and correlate pay and promotions along those measures.
It's the same reason why google products die when they don't reach mega-success (thus products getting killed off in spades recently).
Nobody wants to be doing maintenance on projects started by somebody else - so as soon as the lead visionary leaves for better pastures, that project gets languished, and whoever takes over it cannot use it to generate promotion worthy impact.
I'd go further and bet that those "impact" metrics likely came from a spreadsheet pusher like Schmidt.
Schmidt still takes the gold medal in my estimate for destroying a company he ran by not understanding how the business actually works (as CEO of Novell he decided to screw their channel partners not understanding that the channel relationships were the company's entire moat against Microsoft).
Channel relationships were a boat anchor around Novell's neck. The channel helped Novell grow rapidly in the early days then building a file server and installing an office network took some real technical skill. But then Microsoft released Windows NT and a new generation of hardware came out which allowed any idiot to set up a LAN. Microsoft beat Novell in part by distributing their server software more widely instead of forcing customers to go through channel partners. Of course, it also didn't help that Novell failed to innovate on their products and tried to coast on past success.
Network installers are still used to this day on big network buildouts, simply because office IT departments aren't staffed for that kind of work regardless of how easy the software is.
But it's not like Schmidt had any strategy to bypass the network installers. He just decided to screw them hard so he could make a quarter look good, which lead inevitably to them deciding not to be Novell's low-paid salesforce anymore, which lead rapidly to the vastly premature collapse of the business.
I felt like I could rationalize a lot of Google decisions to kill products, such as Inbox or Reader, but retiring Google Domains really shook my confidence in GCP.
Absolutely staggering. Google Domains complemented GCP perfectly and would make thanks to Cloud DNS a stellar seamless end-to-end integration. I'm in disbelief as to why they removed this crucial feature while at the same time going full in with GCP.
I could understand deprecating Google Domains for B2C, but for B2B?! What went through their heads?
It is actually insane that they did this. I literally stopped recommending people use GCP over this - you can't get started easily because you have to use another platform for the domain. Why even use GCP at all?
It was also incredibly sticky with GSuite. Setting up security and everything was a breeze -- DMARC email records, SMTP, etc all magic -- I truly couldn't believe when this was announced. It was incredibly sad and I still feel upset about it a year later.
Porkbun is my go to registrar now. And I switched my email to Migadu.
Google Cloud Domains is still a thing, and you can purchase domains from the console still. The registrar changed to square space, but it didn't really impact Google Cloud's usage.
Even though I was a Google fanboy, this was a nail in the head. Had a .dev domain which expired and squarespace asked like 6 times more to let me buy it back. Waited till the grace period was over and bought it back for the original price from NameCheap.
> I believe this trend is a result of management culture and performance metrics that attempt to measure "impact", and correlate pay and promotions along those measures.
Meta's performance reviews are also heavily weighted for measurable impact.
I think that the reason why it's beating Alphabet in so many fronts is team culture. Most new product teams are small and have a reasonably flat hierarchy. It's easy to make impact that's both effective and measurable, and the amount of engineering time working in "useless work" is minimised.
Their search isn't even so great any more. They tend to forcefully switch the user input to some generic terms. More than often duckduckgo produces better results for me. I started feeling comfortable using duckduckgo instead of google.
You are just ignorant, Google search is 57% of their total revenue, remaining advertisement is just 19%, people really overestimate how much Google ad network is worth when its just 9.2% of their revenue and YouTube is just 10%. In fact if you remove search Google makes more money from selling subscriptions and services than they do from ads, so they would no longer be an ad company!
Facebook makes much more money than 19% of Google.
Edit: And I wonder why I got downvoted for being right, many here just blindly believe that Google gets their money from ads on third party sites when most of it comes from search.
But I guess your second post is right, it isn't close, Facebook would be much bigger.
> Google ad manager is the cornerstone of pretty much all internet advertising.
That money is a part of Googles revenue, so there isn't as much there as you might think. Maybe you mean there is quantity there but advertisers doesn't pay as much for it as they would for Facebook or Google search ads, then yeah that is possible, but normally in a market segment you count big players by revenue and not numbers sold.
Fuchsia is part of the problem not the solution. It was the retirement home to keep a bunch of smart people busy who didn't have any rush to ship anything real for as long as possible. Architecture astronaut porn to the max.
On the other hand there's zero desire at Google to try and build the best phone. They have conceded that to Apple a long time ago.
No desire to be number one in Cloud. They are fighting for number 3 spot.
They are the Xerox of AI with a full stack solution. If they had the drive they would be HYPING THE SHIT out of TPU and making it the best solution to run all the real workloads people have.
Guess what. On average a Googler even still today with low bar to entry is probably 10 IQ points higher than average Amazon or Microsoft or Apple. But no; no drive.
I hadn't heard the term "architecture astronaut porn" before. Seems highly accurate in this context and wondered if you could point me to other resources on this term?
I know nothing about what makes an industry succeed or fail, and also nothing about web tech, but working in the field I can comment on:
> tensorflow looks like currently loosing to pytorch - seems like google got bored and more development is for JAX, Keras wrapper
Well, TensorFlow doesn't "look like currently losing", it has already lost since a long time. I haven't seen a decent paper release code in TensorFlow in years, and all the references I see to TF online are job posts from "older" companies (to the point that, if you are looking for a job in data science, seeing TF mentioned in the job post is kind of a red flag of a place you don't want to be).
That said, I am quite certain that this has only a small impact on why Google is losing terrain, and even on why it is behind in AI (which is also debatable: narrative aside, Gemini is not that much lacking behind competitors). Certainly if TensorFlow + TPUs turned out to be better than PyTorch + GPUs they would have had a lead to start from, but if that was so important, Meta or NVIDIA would have created the first LLM, not OpenAI.
Simply, sometimes stuff happens, you can't predict it all.
> Considering that in Meta I heard (from friends) it is/was easier to work remotely and even easier to get into (they are less leetcode oriented)
This is a weird take. It might pertain to 2020 during Covid when Meta encouraged remote working, but it's not like that anymore. Interviews were always leetcode focused so nothing changed there.
Meta is an interesting comparison because the performance culture has historically been way tougher than Google (sans Covid years). It's quite a shock for a lot of Xooglers who join. One Xoogler at Meta who told me they only used to work 4-6 hours a day at Google. That is basically impossible in Meta - you have to constantly prove your value and the perf bar is high.
I think Google is changing now though and I wouldn't be surprised at all to see them adopt Meta/Amazon style perf cultures in the future.
Source: I work at Meta, and have close friends at Google.
Compared to at least as bad, or even worse ‘toll’, for dramatically less pay in literally every other industry and 95% of every other companies in this one? Bwaha.
Have you ever worked in another industry? Retail? Healthcare? Trades? Public safety?
Healthcare? Public safety? Most construction work?
All of them have the same thing going on. Hell, healthcare has even formalized it in some cases into 24+ hr shifts for residents.
In all these industries, there are areas where it isn’t a thing, and areas where it is absolutely a thing. But they do (often) get paid overtime. Not residents though.
Meta is less leetcode oriented? I'm shocked to read this. Meta is the poster child for leetcode style interviews. Meta requires you to solve 2 leetcode style questions in 35 mins (out of 45 mins - first 5 mins for initial pleasantries, last 5 mins for asking questions). For each question, you're required to (based on the signals they look for) ask clarifying questions, present a solution to the interviewer, get buy-in, code, verify with test cases - all this in 17.5 mins/question. Go figure! :-)
Tensorflow losing has nothing to do with Google getting bored -- it's vice versa.
Tensorflow is a symbolic framework, which is less intuitive to work with for most people than the Pytorch. Not to mention the errors Tensorflow generates are more annoying to debug (again more an issue with the fact that it's symbolic than any lack of effort on part of Google)
Google tried to fix it by introducing an eager mode in Tensorflow but by then it was too late.
> Google tried to fix it by introducing an eager mode in Tensorflow but by then it was too late.
And the fix was "new major version with a fundamentally different programming paradigm"
But it turns out when your users are irritated with your product, and you tell them to change to a fundamentally different programming paradigm, the new programming paradigm they change to might not be yours.
Intuitive has nothing to do with it. Developers will tend to prefer things that make their lives easier. Debuggability is a huge part of that. Tf 1.0 having a static execution graph was a major pain. No wonder people switched to PyTorch and didn’t look back.
All of those examples (except maybe tensorflow, I don't know enough to say for sure) are interesting because they highlight a more "googly" approach to the problems.
Typescript is a superset of javascript, Dart isn't (though it transpiles).
Flutter implements its own widgets, react native controls native widgets.
Dart, angular don't matter in business. They are just "passionate" projects. Google's culture of having multiple teams trying to build and scrape and never commit to a product is a big problem. They aren't slow to adopt to trends but they always seem to "overthink" about products. So when they lose the edge they close the door on the product. In the case of Google Meet, Good Lord, they renamed at least six times before they finally settled.
I recently did interviews with Meta, Google, Amazon, Apple, Nvidia... Meta leetcoded the ** out of me. They were insisting on 3 days in office. Definitely the hardest interviews out of the bunch. Never got around to answering Netflix recruiter to get the tradition "N" in FAANG out of the way, I took the first offer I got, I was so tired of interviewing.
>dart lost to typescript on web
>angular lost to react
>tensorflow looks like currently loosing to pytorch - seems like google got bored and more development is for JAX, Keras wrapper
Google seems to have a deep-seated distrust for programming language theory, as well as a deep-seated distrust for its users. This combination produces awkward software and APIs that ignore modern PLT and take a "Google knows best" approach.
As someone who's been working with them a lot at the moment I'm going to say that the problems that google has are not anything to do with tensorflow or programming language theory and everything to do with the fact that the place is absolutely jam-packed with MBAs and other ex-McKinsey-type professional meeting attendees. This is really apparent if you are reasonably senior in an enterprise and deal with google as a vendor.
Every meeting I attend with them has one or two engineers[1] struggling to breathe because there are at least 6 or 7 sales people, relationship managers or other non-doing non-technical middle-management spreadsheet jockeys stealing all the oxygen in the room.[2]
I can only imagine how terrible it is to work there given that these folks have all the power internally. I've genuinely never dealt with an organization that seems this bad.
[1] who are usually pretty good.
[2] It's an internal joke at my enterprise how every meeting another new person from google shows up introduces themselves as head of some other microscopic facet of the corporate relationship.
As someone two years into working at Google this resonates hard and has absolutely been my experience as an engineer there too. It's certainly not the culture I was expecting for sure.
Longer-time Googler here. We went blindingly fast from "A few engineers decide to use the most powerful computing cluster in the world to make a meme generator because it would be a cool project" to "Let's have a sync meeting with 15 people, including five managers, to discuss buying a $20,000 test instrument, which will need to be approved by four directors, three of whom are OOO for the next two months."
Something changed recently though because yes there are those MBA types but the engineer in the room used to be not a sales type person. Now (earlier this year, just before IO), even the engineers are becoming more like salespeople in my experience. My guess is there is pressure for everyone to make more money somehow?
this is the end result of all publicly traded companies. eventually you'll hit Sears-level of selling where the CEO just straight up puts departments in conflict with each other until the whole thing just falls apart.
If I had to guess you're probably an Enterprise customer of Google cloud? I'm not sure that's a representative take.. of course there's going to be a lot of sales and relationship manager types because you're a customer of their Enterprise offering.. internally Google's completely different
- dart lost to typescript on web
- angular lost to react
- tensorflow looks like currently loosing to pytorch - seems like google got bored and more development is for JAX, Keras wrapper [0]
- IMHO flutter will loose with react-native or kotlin compose multiplatform - compare github insights for details
Meta on the other hand kickstarted open source Llama community. In this situation it's hard to bet on Gemmini or Gemma as 3rd party developer considering google projects kill records. The only project they were really to bet on and invest for the long run without getting tired early on was Chrome and Android.
[0] https://trends.google.com/trends/explore?date=today%205-y&q=...