Most active commenters
  • dilyevsky(6)
  • sjwhevvvvvsj(4)
  • thanksgiving(4)
  • linkgoron(3)
  • reissbaker(3)

←back to thread

FAQ on Leaving Google

(social.clawhammer.net)
462 points mrled | 56 comments | 17 Jan 24 22:12 UTC | HN request time: 2.324s | source | bottom
Show context
thrtythreeforty ◴[17 Jan 24 23:40 UTC] No.39035233[source]▶
The author also published [1] an email he wrote at the beginning of his tenure. It is amazing how alien and out of place early Google sounds in today's corporate environment. They have completely eroded the perception that Google is this kind of place:

> Google is the opposite: it's like a giant grad-school. Half the programmers have PhD's, and everyone treats the place like a giant research playground [...] Every once in a while, a manager skims over the bubbling activity, looking for products to "reap" from the creative harvest. The programmers completely drive the company, it's really amazing. I kept waiting for people to walk up to me and ask me if I had declared my major yet. They not only encourage personal experimentation and innovation, they demand it. Every programmer is required to spend 20% of their time working on random personal projects. If you get overloaded by a crisis, then that 20% personal time accrues anyway. Nearly every Google technology you know (maps, earth, gmail) started out as somebody's 20% project, I think.

Even if this was only half-true back then, there's very little you could do to convince me that it's true at all now. This culture and the public perception of it has been squandered.

[1]: https://social.clawhammer.net/blog/posts/2005-09-25-FirstWee...

replies(6): >>39035428 #>>39035523 #>>39035569 #>>39035617 #>>39035738 #>>39046460 #
1. sjwhevvvvvsj ◴[18 Jan 24 00:29 UTC] No.39035738[source]▶
Both Maps and Earth were acquisitions. MOST Google products are.

The only two real big success products to come from Google that are still around are Search and Gmail. Maybe you can count Scholar but it’s really just a type of search.

Workspace was assembled from various acquisitions, YouTube they bought, Cloud is just a Jack Ma-esque “copy whatever Bezos is doing” initiative.

Most home grown Google products have either failed or been killed in the cradle. G+, Stadia, etc etc

20% was always a myth.

replies(6): >>39035852 #>>39036708 #>>39036792 #>>39037838 #>>39038129 #>>39045661 #
2. United857 ◴[18 Jan 24 00:40 UTC] No.39035852[source]▶
Google Chrome seems like a success as well.
replies(2): >>39035950 #>>39037822 #
3. linkgoron ◴[18 Jan 24 00:50 UTC] No.39035950[source]▶
Also forked from something Apple made (Webkit)
replies(4): >>39036019 #>>39036418 #>>39036726 #>>39037459 #
4. zbowling ◴[18 Jan 24 00:57 UTC] No.39036019{3}[source]▶
1) using that test, Apple didn't make webkit either. It's a fork of KHTML and why everything still uses LGPLv2 2) very little of what goes into making a browser successful is just the render. In WebKit and now Blink make up only a small percentage of the total browser.
replies(1): >>39036196 #
5. linkgoron ◴[18 Jan 24 01:16 UTC] No.39036196{4}[source]▶
Chrome/Chromium was developed for quite a while using Webkit. Chromium was created in 2008 and only after Google had already captured a third of the browser market share (according to Statista) did they fork it (April 2013).

The fact that basically all of the big companies (Microsoft, Google, Apple) use Webkit or Chromium shows that it's very difficult to build and maintain one successfully IMO. I think that Mozilla are essentially the only ones developing something that's somewhat competitive, not to mention that most smaller companies (e.g. Opera, Brave, Vivaldi, Island etc.) all use Chromium.

I'm not saying that it's easy to succeed with a product even after you've bought it, or started it from a fork (see less successful Chromium/Webkit forks). I'm just saying that it was not something built from the ground-up in Google. For example, v8 was and really changed a lot of things in the JavaScript world including Node, Deno etc.

replies(2): >>39036598 #>>39038836 #
6. sjwhevvvvvsj ◴[18 Jan 24 01:42 UTC] No.39036418{3}[source]▶
And Apple forked WebKit from KDE!
7. reissbaker ◴[18 Jan 24 02:06 UTC] No.39036598{5}[source]▶
I think v8 and the multi-process model were the big differentiators of Chrome when it first launched, and how it originally got marketshare! Regardless, I think "ground-up" building isn't a great way to measure product building; after all, macOS is "just" a BSD fork, as others have pointed out Webkit was originally a KHTML fork, etc. And just about any web product runs on Linux and is effectively a wrapper around libc, which wasn't ground-up built by any modern tech co.
replies(1): >>39038370 #
8. dilyevsky ◴[18 Jan 24 02:20 UTC] No.39036708[source]▶
Android was developed entirely at google (and redone midway after iphone came out) despite being originally an acquisition. Youtube basically just sold userbase + content. Chrome. Waymo. AppEngine precedes ec2 and heroku by some time. Most of hashicorp products (and dozen other startups) are basically copies of what google had internally.

The theory that google hasn’t birthed any original products just doesnt hold any water

replies(5): >>39036765 #>>39037748 #>>39038312 #>>39038516 #>>39038981 #
9. dilyevsky ◴[18 Jan 24 02:22 UTC] No.39036726{3}[source]▶
That’s like saying os x was just a fork of bsd
10. commandersaki ◴[18 Jan 24 02:26 UTC] No.39036765[source]▶
Wasn't the idea of Android basically the acquisition of Danger Inc.
replies(3): >>39036832 #>>39036919 #>>39037365 #
11. lokar ◴[18 Jan 24 02:29 UTC] No.39036792[source]▶
20% was very real, I saw it many times

Googles main successful product was amazing infrastructure. Lots of real innovation. It enabled massive scale of everything else, including acquisitions. YouTube was about to hit the wall when they got acquired.

replies(2): >>39037434 #>>39037617 #
12. dilyevsky ◴[18 Jan 24 02:33 UTC] No.39036832{3}[source]▶
Kind of - danger was bought by msft, then everyone left and joined Android/Google. I think their original plan more akin to those chinese all-in-one apps
replies(1): >>39037917 #
13. eigen ◴[18 Jan 24 02:43 UTC] No.39036919{3}[source]▶
I believe it was Android Inc. that Google purchased. Danger was a previous company founded by Andy Rubin and others.

https://web.archive.org/web/20110205190729/http://www.busine...

14. shermantanktop ◴[18 Jan 24 03:29 UTC] No.39037365{3}[source]▶
This kind of lineage is interesting, but I don’t give large amounts of credit-for-success to a company that failed at what they were trying to do, or gave up and sold themselves off. How much of why Android is huge today could really be attributed to Danger? Not too much, in my book.

Can we really say that Danger could have accomplished the same thing? I was in the carrier industry at that time and Danger was just another handset company.

replies(1): >>39037823 #
15. cmrdporcupine ◴[18 Jan 24 03:36 UTC] No.39037434[source]▶
By the time I got to Google in 2012, 20% seemed dead to me. If it had any meaning, it just meant "you can do some extra work on something management approves", not "I have a wild idea and want to go off and try it in my 20%" (as I naively understood it before coming there.)

It was already the case, at that point, that 20% really just meant doing more of what Google was already doing.

But maybe I just didn't know the right people or have the right connections or status.

replies(1): >>39037544 #
16. cmrdporcupine ◴[18 Jan 24 03:38 UTC] No.39037459{3}[source]▶
The original genius in Chrome was not the renderer built out of webkit. It was:

1. The V8 JavaScript engine, which blew away everything else. 2. The sandboxed, multiprocess, threading model.

Those were the two things emphasized in the original Chrome "comic" at launch, if I recall:

https://www.google.com/googlebooks/chrome/

replies(1): >>39037938 #
17. lokar ◴[18 Jan 24 03:49 UTC] No.39037544{3}[source]▶
Yes, by then it was org dependent. Some were not supportive, but some were.

2005 was still very open

18. sjwhevvvvvsj ◴[18 Jan 24 03:59 UTC] No.39037617[source]▶
100% - Google IS SRE. My favorite group of people to work with when I was there, a true honor, just amazing infra.

But the people “leading” the place are trash.

19. gretch ◴[18 Jan 24 04:14 UTC] No.39037748[source]▶
YouTube was founded in 2005, and then sold to Google in 2006.

Then it was run under Google from 2006 to 2023.

Does anyone remember what 2005 looked like at all?

But people really like the narrative that Google couldn’t make a YouTube

replies(3): >>39037907 #>>39038022 #>>39038749 #
20. thanksgiving ◴[18 Jan 24 04:23 UTC] No.39037822[source]▶
My guess is Google Chrome spends well over a billion dollars a year and comes up with ridiculous rules like this https://developer.chrome.com/blog/autoplay#media_engagement_... because Google Chrome is Google first and a web browser second.

It will collapse under its own weight if Google stops spending billions of dollars on it every year.

replies(2): >>39037927 #>>39045061 #
21. klooney ◴[18 Jan 24 04:23 UTC] No.39037823{4}[source]▶
Android's whole design is very Danger though- the Java userspace, Binder RPC stuff.
22. binkHN ◴[18 Jan 24 04:25 UTC] No.39037838[source]▶
Stadia was amazing.
23. throwaway2037 ◴[18 Jan 24 04:37 UTC] No.39037907{3}[source]▶
And, there is no way that YouTube could survive on its own. The last mile bandwidth problem required a Google-sized company to help them solve it. This is usually overlooked.
replies(1): >>39038170 #
24. thanksgiving ◴[18 Jan 24 04:38 UTC] No.39037917{4}[source]▶
I thought the original idea was instead of having to download and run random JAR files for random Nokia or Erickson phones, wouldn’t it be nice to have an open handset alliance that would allow developers like Google to write their applications only once and it would work on all phones running android…
25. throwaway2037 ◴[18 Jan 24 04:40 UTC] No.39037927{3}[source]▶
This is an interesting point. Can anyone from the inside estimate the annual budget for Chrome? A billion sounds like a lot. That implies: 1b / 250k = 4,000 (expensive) developers. I guess at least 1,000 well-paid people are involved, so hundreds of millions sounds more likely.
replies(4): >>39037986 #>>39039255 #>>39040989 #>>39040999 #
26. thanksgiving ◴[18 Jan 24 04:42 UTC] No.39037938{4}[source]▶
Kind of easy to forget the true innovation of Google chrome these days. I will try to remember this again any time I see an aww snap on my web browser because it would have been all tabs all windows dead at once before Google chrome.

Firefox only declared it completed electrolysis in 2018, nearly a decade after this comic.

27. thanksgiving ◴[18 Jan 24 04:48 UTC] No.39037986{4}[source]▶
There is also advertising and cross promotion. I am also counting the opportunity cost of ads not sold because the spot went to Google Chrome.

Disclaimer: I’ve never worked at Google and have no insider information.

28. xnx ◴[18 Jan 24 04:54 UTC] No.39038022{3}[source]▶
I remember Google Videos being better than YouTube at the time, but IIRC it didn't have the amount of pirate content that initially made YouTube popular.
replies(1): >>39047090 #
29. RestlessMind ◴[18 Jan 24 05:08 UTC] No.39038129[source]▶
> The only two real big success products to come from Google that are still around are Search and Gmail.

Chrome. Photos. Hadoop. Kubernetes. Brain. Spanner. T in GPT (Transformers). And lots more. Google's real contribution was internet scale systems and how to run them reliably.

replies(3): >>39039117 #>>39042486 #>>39061785 #
30. dilyevsky ◴[18 Jan 24 05:14 UTC] No.39038170{4}[source]▶
Legend has it they had like weeks of runway left and didn’t pay any of their bills once it looked like the deal was going to close
31. choppaface ◴[18 Jan 24 05:37 UTC] No.39038312[source]▶
Not so much that Google hasn't birthed _any_ original products but rather that their customer service is abysmal and they've consistently shown poor long-term commitment to the end user, or worse, e.g. Reader, Nest, Fitbit, even Tensorflow is dead. The theory is that Google makes it needlessly hard for product people to innovate there, and the evidence is in Google's outsized insularity and coddling of technical projects that end up mostly for internal entertainment.
32. edgyquant ◴[18 Jan 24 05:46 UTC] No.39038370{6}[source]▶
MacOS is not a fork of BSD but uses some of its use land. I think it’s considered a BSD because of that, but the kernel and graphics libraries are all Apple.
replies(1): >>39040199 #
33. yjftsjthsd-h ◴[18 Jan 24 06:09 UTC] No.39038516[source]▶
> Youtube basically just sold userbase + content.

Those are kinda important parts. Like, to the point that if they'd homegrown "GVideos" I bet it would have failed.

> Chrome.

Which was a WebKit wrapper - explicitly just the browser chrome.

replies(2): >>39038861 #>>39038933 #
34. Scoring6931 ◴[18 Jan 24 06:46 UTC] No.39038749{3}[source]▶
They had a YouTube. It was called Google Video, and it went nowhere.
35. taylortbb ◴[18 Jan 24 06:59 UTC] No.39038836{5}[source]▶
> Chrome/Chromium was developed for quite a while using Webkit. Chromium was created in 2008 and only after Google had already captured a third of the browser market share (according to Statista) did they fork it (April 2013).

I think you missed the point, there's two forks in the history of Blink (Chromium). Yes, Blink is a fork of WebKit, but WebKit is a fork of KHTML. So it's not like it originated at Apple either, it originated at KDE.

replies(1): >>39039749 #
36. zilti ◴[18 Jan 24 07:03 UTC] No.39038861{3}[source]▶
They did have Google Videos before they bought Youtube, and integrated it a while after.
replies(1): >>39039200 #
37. dilyevsky ◴[18 Jan 24 07:17 UTC] No.39038933{3}[source]▶
> Those are kinda important parts. Like, to the point that if they'd homegrown "GVideos" I bet it would have failed.

The YouTube product which is their creator economy that exists today didn't back then. In fact, I'm pretty sure original team would've run out of money soon.

> Which was a WebKit wrapper - explicitly just the browser chrome.

And Docker is "a wrapper" around Linux Cgroups. So? It was a unique product with instant market fit - "fast browser without the UI clutter and with sandbox'ed tabs".

38. billjings ◴[18 Jan 24 07:25 UTC] No.39038981[source]▶
This is a really off base characterization of Android within Google.

Chet Haase wrote a book on those years, and while it is clear that Google gave them rocket fuel to meet their ambitions, their company culture was wildly different from the rest of Google. Shipping code on Android would not have passed muster for anyone at mainline Google; the process and standards were utterly alien from one another.

There is no way Android happens without the acquisition.

replies(2): >>39039785 #>>39046819 #
39. pcdevils ◴[18 Jan 24 07:48 UTC] No.39039117[source]▶
Google photos was spawned from Picasa, which they bought https://www.wsj.com/articles/SB109121493116979168
replies(1): >>39041001 #
40. ponderings ◴[18 Jan 24 08:02 UTC] No.39039200{4}[source]▶
Right, it was google video (that was just a bunch of users and content) that was "merged" into youtube. As usual they didn't bother redirecting the url's. Just let all of those hundreds of millions of links rot. What an opportunity to ruin an unimaginable number of threads and blog posts.

I'm trying to picture a white board with someone drawing up a plan how to destroy everything and take over.

Woah, video replies, we have to remove those. Threaded conversations under videos? Lets make them into an unbearable mess and make it as hard as possible for anyone to have a conversation. We can put it under history! ha-ha good one! Wait, we could suck everyone into a vacuum and have them all watch the same videos? ~ Excellent idea!

Creative company indeed

41. I_AM_A_SMURF ◴[18 Jan 24 08:11 UTC] No.39039255{4}[source]▶
Mozilla at least at one point had 1,000 and is and always was chronically underfunded. 4,000 seems about reasonable. Keep in mind it's not 4,000 engineers, it's PMs, managers, UX, Infra, there's a lot more to software development that just line engineers.
42. linkgoron ◴[18 Jan 24 09:27 UTC] No.39039749{6}[source]▶
I did not miss the point, I just don't see why it's relevant. This isn't a thread about Apple's products and their success. The fact that Apple started from KHTML is not really relevant. However, it's clear that at the beginning Google was very dependent on Webkit and Apple, and there's a good reason why it took them five years of gaining development expertise and market share before forking Webkit.

I've already stated that Chrome's success is not just because that it was forked from Webkit (e.g. v8, and other things that people mentioned here as well), but it was a huge jumpstart for them, and it would've taken them much longer to get a leading browser without it. e.g. Microsoft basically gave up on developing their own engine after failing with IE and the original Edge - and are now also based on Chromium.

Chrome is (IMO) much better than Safari, Maps is (IMO) a great product, Youtube is a a huge success and much bigger than it was when they bought it (homegrown Google Video failed), Android was also essentially an acquihire, as others have mentioned (using a lot of Google's resources) and is hugely successful. It doesn't change the fact that most existing Google products today are acquisitions that they improved, and not home-grown products from the "20% do your own thing" era - which is what the original comment talked about.

43. versteegen ◴[18 Jan 24 09:34 UTC] No.39039785{3}[source]▶
Yeah, when I first looked at the shocking source code for bionic (Android's libc) to figure out why my code wasn't working I couldn't believe it was written by Google. It wasn't really. (Nor did they (initially) borrow from any of the high-quality open source libcs out there.)
44. reissbaker ◴[18 Jan 24 10:37 UTC] No.39040199{7}[source]▶
The kernel isn't all Apple, it's a fork of the open-source Mach kernel developed at CMU (which was a replacement BSD kernel). "Ground up" just isn't real!

The graphics libraries are definitely more custom... Although in total fairness they're not entirely ground-up Apple either; Quartz was based on Display PostScript, which was acquired from NeXT, and which NeXT built in collaboration with Adobe based on Adobe's earlier work on PostScript. It's obviously true Apple's done a lot of work since then (e.g. Metal), but in that case, so has Google since forking Webkit.

replies(1): >>39051356 #
45. gowld ◴[18 Jan 24 12:39 UTC] No.39040989{4}[source]▶
Google engineers cost closer to $500K/head all in.
46. room500 ◴[18 Jan 24 12:39 UTC] No.39040999{4}[source]▶
250k is conservative for employee cost. A staff engineer at Google can reach 1 MM total comp. And add in all the overhead a company has (real estate, free food, perks, taxes, etc)

500k-700k is a little more realistic. 1500 employees across all domains (engineering, marketing, product management, customer service, etc) isn’t a huge number

47. gowld ◴[18 Jan 24 12:40 UTC] No.39041001{3}[source]▶
Photos totally replaced Picasa, including, you know, replacing a local desktop app with a web/cloud app.
48. murki ◴[18 Jan 24 15:01 UTC] No.39042486[source]▶
Google Photos came from the acquisition of Flock (via Bump) which was a mobile photo organizing and sharing app https://techcrunch.com/2013/12/31/google-to-close-bump-and-f...
49. hardwaregeek ◴[18 Jan 24 17:59 UTC] No.39045061{3}[source]▶
A billion dollars is cheap considering it saves Google from having to pay Apple or Mozilla more money to stay the default search engine. Google gives Apple 10 billion a year just in traffic acquisition payments.
50. gniv ◴[18 Jan 24 18:33 UTC] No.39045661[source]▶
> Both Maps and Earth were acquisitions.

This is technically true, but Maps before Google was nothing like what you see now. Most of the innovation happened at Google. Earth hasn't changed that much, but it's not really the killer app that Maps is.

51. dilyevsky ◴[18 Jan 24 20:01 UTC] No.39046819{3}[source]▶
A lot of teams at google had culture differences. And imho it was always more about andy's ego more than anything.
52. sjwhevvvvvsj ◴[18 Jan 24 20:20 UTC] No.39047090{4}[source]▶
This was borne out in lawsuits. The YouTube tech wasn’t better, GVideo was superior but disnt have the same buzz or content, it was the pirated content Google wanted. The emails are public record.
53. astrange ◴[19 Jan 24 03:47 UTC] No.39051356{8}[source]▶
Quartz isn't based on DPS - CoreGraphics drawing commands are similar to PDF but the window management was always all pixel based - and IIRC DPS was almost all Adobe and NeXT didn't even have source for it.
replies(1): >>39052335 #
54. reissbaker ◴[19 Jan 24 06:50 UTC] No.39052335{9}[source]▶
Quartz was originally based on PDF and was marketed early on as "Display PDF" (a reference to being a more advanced version of "Display PostScript"): https://archive.arstechnica.com/reviews/1q00/macos-x-gui/mac...
replies(1): >>39054510 #
55. astrange ◴[19 Jan 24 12:08 UTC] No.39054510{10}[source]▶
Well, it wasn't really based on PDF to the point it stores PDFs in memory or anything. Like I said, it just has compatible drawing commands.

It's turned out that vector images aren't that performant or easy to deal with for various reasons, although they are still used in some places.

56. dgacmu ◴[19 Jan 24 21:55 UTC] No.39061785[source]▶
Hadoop didn't come from Google, it came from more like Yahoo, who wanted an open source clone of Google's mapreduce framework. (Yahoo was paying Doug's salary, at least). MapReduce was deprecated inside Google several years ago in favor of Flume (think "apache beam" in the open source world).

I point this out because google has had a lot of innovations that aren't necessarily now _products_.