Hacker News Reader

Comment by mabbo | Hacker News Reader

Many years ago, I interned at Google and a very wise mentor of mine explained Google to me in a way that has stuck with me.

"Google discovered a hose that money poured out of. It's called 'online advertising'. All that we do now is find ways to make that hose go faster, and desperately search for another hose."

Why does Google develop 3+ different OS's? In case one of them is a money hose. Why has Google created thousands of products, only to cancel most of them once people fell in love? Because they decided that they weren't money hoses.

View on HN

> All that we do now is find ways to make that hose go faster, and desperately search for another hose.

There’s one other thing Google does a lot of: try to foresee and actively forestall how someone else might cut-off or divert the flow to the existing hose. While both Chrome and Android evolved in directions that explored potential new hoses, they were clearly both initially directed at protecting the existing hose from being choked off or diverted by monopolies on the client side; Android was purchased and engineered to prevent Apple gaining the kind of dominance in mobile to do that, Chrome and Chrome OS against the desktop Windows monopoly and IE.

Google+ was deployed against Facebook and abandoned once it was clear that Facebook’s position, while making it something of a competitor for ad dollars, wasn’t, even without a competitive Google alternative, an existential threat to Google’s money hose.

This is the big one that people forget. Android itself isn't there to force tracking and marketing on users. It's there to prevent somebody else from controlling mobile devices and being a gatekeeper between Google's services and users. As long as nobody else can prevent Google from reaching users, Google doesn't care that people can run Android without Google services.

In practice, poor development practices and incompetent product management has left some bits hard to change. The search provider used to be hardcoded with no way for the user to override it. That only changed when some early Android phones began shipping with Bing as the default. To this day, it is impossible to allow any non-system app store to update apps in the background. This hasn't changed only because Amazon, Samsung, and many other OEMs failed to get developer traction with their own app stores. Now that there are antitrust rumblings, Google has hinted they will finally fix this, but once again, this isn't a threat to the core Android strategy of removing gatekeepers.

> Android itself isn't there to force tracking and marketing on users. It's there to prevent somebody else from controlling mobile devices and being a gatekeeper between Google's services and users.

It can be, and is, both.

If that were the case, it would require more data collection than competitors and force ads. Instead, you can trivially get your location on Android devices (even those sold by Google) without telling Google. You cannot on iOS. You can install apps on your Android device without telling Google. You can develop apps for your own device without telling Google. To do any of these things on iOS requires exploiting a vulnerability.

Similarly, Google doesn't give itself a privileged separate settings for marketing data collection that ignore global settings. https://support.apple.com/en-us/HT202074

If the point of Android were data collection and marketing, Google would abuse configurability of data collection at least as much as Apple does, but it does not, not even on devices that Google itself sells.

Google Play Services is essential for most features a user would expect in a modern mobile phone. For running bank apps (SafetyNet), many maps apps or apps that include maps (eg Uber, Tinder - due to Google Maps API), push notifications, etc. It's absolutely used to lock-in users to the Google ecosystem and discourage them from degoogling.

Those are services that can be provided by others on any Android device, including those sold by Google. The fact that others aren't providing those services doesn't mean Android itself is meant to lock in users (or to collect data / serve ads), only that others haven't found a reason to provide those services. Google also provides many of those services on other OSes it does not develop (e.g., Chrome push notifications and Widevine).

nit: Android was purchased before the iPhone was announced.

Yep, the reality is that it was more likely a defensive play against Microsoft. At the time (mid-2005) it seemed most likely that Windows Mobile would eventually take over, and Android's strategy was taken from the Windows Desktop playbook.

Google had to postpone Android phones once iPhone launched because Google realized how badly they'd misjudged smartphones, particularly on-screen keyboards & multi-touch (& thinking that their primary competitor was Microsoft and Nokia & not Apple). It's actually a testament to the leadership that they managed to recognize that Apple was the 800lb gorilla here so quickly (which clearly wasn't obvious to other competitors in the space).

https://venturebeat.com/2008/08/07/first-android-phones-push...

> which clearly wasn't obvious to other competitors in the space

I had a family friend that was a tech writer in 2007 or so. He proudly mocked my interest in the first iPhone. Pulling me aside and showing me how his blackberry had a mini trackball and a mouse cursor and just how precise this all was.

It’s obviously funny now, but yea I remember a lot of the “smart money” was against the modern smartphone at the time.

Also, Google's internal politics are kind of broken. It's my understanding that a lot of the product launches and cancellations are a consequence of an internal culture that rewards launching new products, but not maintaining or improving them.

Neither OP or GP have it right - Fuchsia isn't a "3rd OS", it's a kernel, like Linux, except MIT. Maintenance is most certainly rewarded, at least in the 5 years I've been at Google

Eh. If you've been at Google for 5 years then you should recognize Fuchsia as another one of the "retention projects" where you get paid to twiddle away at something cool in the corner and your only real deliverable is not going to work for Facebook or Apple.

Sometimes they stumble upon something another PA needs, but often it just ends up as a patent or research paper.

Why would Google pay employees to not go to Apple?

The only competition for the giants are the other giants.

Who can divert from Google's massive ad income funnel?

Apple.

Facebook.

(Also maybe several foreign governments' antitrust cases.)

An engineer on your payroll is an engineer not on the competition's payroll.

Because Google has infinite money, the execs just hire people so they can admire the zoo of smart people they've built. Why do they need Vint Cerf, Guido, and Rob Pike? Why are they on the MIDI board? Because there's no real reason not to.

Guido hasn't been at Google since 2013

I don't have an opinion either way on your position here, other than a fact check.

Guido van Rossum (creator of Python) is a Distinguished Engineer at Microsoft. Before that, he worked at Dropbox.

Before that, he was at Google. Failing to retain an employee doesn't invalidate that Google would rather they didn't leave.

I get the quarterly check-ins from the "alumni recruiting" team, and can assume Guido does too.

Facebook or Apple? Not Amazon or Microsoft?

Actually, it is a 3rd OS named Fuchsia, which uses a custom kernel called Zircon which is MIT.

"Zircon is the core platform that powers Fuchsia. Zircon is composed of a kernel (source in /zircon/kernel) as well as a small set of userspace services, drivers, and libraries (source in /zircon/system/) necessary for the system to boot, talk to hardware, load userspace processes and run them, etc. Fuchsia builds a much larger OS on top of this foundation."

- Fuchsia Docs, https://fuchsia.dev/fuchsia-src/concepts/kernel

I suspect that this is a consequence of the above problem. Something that's pretty successful and growing at a pretty good clip is not a giant spigot of money, and therefore it's effectively useless to Google. Maintaining it requires programmers that could be better put to searching for the next spigot.

This is also typical of mismanaged governments:

Always building new infrastructure, and not maintaining the existing one.

As I understand it, higher level promotions require a portfolio be compiled and submitted to an anonymous review board. A successful product launch is one of the most valuable additions to these packets.

This has two logistical consequences: 1 - it's more valuable to launch a new product that it is to successfully maintain/grow a current one. 2 - the person running these new projects tends to get promoted away, leaving it rudderless.

Google could change all this overnight by restructuring what these review boards prioritize.

If a solution is that simple, why hasn't it been done? Also, the head of Android has become the CEO, and Android is still not rudderless.

Because the system works? Why fix what isn't broken?

It works for some things but they've kind of painted themselves into a corner with cloud services, many companies do not trust that GCP will not just disappear without a trace someday.

Identifying a good change and actually implementing it in a massive organization are two wildly different things.

People always say this on HN, and while it’s definitely a viable explanation for why Google so often cans projects that are well-loved, I’m finding it harder to believe the more I think about it. If all of us on the outside are so aware of this fact about Google’s culture, surely someone with sway has heard about it, and taken steps to fix it. I’m sure Google are well aware of the perception that they’re always getting rid of products, and would like to fix that. So if the explanation was just that the culture doesn’t reward maintaining existing projects, I’m sure we would have heard something about a corporate incentive to maintain stuff.

I hate the word, but if there's one thing google has it's analytics - data. They can tell if something is growing or dying. I'm guessing they could even tell youtube was growing before they bought it. But seeing an upward trend seems easier than seeing a flat line and deciding it can't be turned into an upward trend.

> So if the explanation was just that the culture doesn’t reward maintaining existing projects, I’m sure we would have heard something about a corporate incentive to maintain stuff.

So, two things:

1) Changing culture is really hard, especially at a company the size of Google, and

2) Even if they could change the culture, it's possible they simply don't care or they believe the benefits outweigh the cost.

How about this explanation instead: Google doesn't actually have a habit of canceling projects that are well-loved?

If you don't believe it, try selecting 10 of their supposedly killed products at random. Had you honestly heard of them ever before? Do they look like something that would have had a significant user base? Hell, do they even look like they were actual products?

> product launches and cancellations are a consequence of an internal culture that rewards launching new products, but not maintaining or improving them

This is very stale information at this point. Product launches are now pretty heavily gated, and promotion criteria are far less based on launches than on overall impact, including but not limited to improving/maintaining products.

(Googler but interpretation is my own)

I think that depends where you are. There's a lot of lip-service being paid, but from what I've seen, it's mostly translated to talking about _landings_ rather than _launchings_, which isn't much of a change.

(Also a Googler.)

I think landings allows for a much broader scope of accomplishments. For example, if refactor an entire system for demonstrably better maintainability and efficiency, that's a landing, but not a launch.

People say this all the time about every company. For many this is true, when seen from the narrow perspective of an individual. However for the only metric that people care about (stock price and profit) Google’s system is incredibly successful. Higher ups will not dictate changes like this if the company is successful. Culture change at the macro level only happens when the company is spiraling at the macro level.

This seems like a pretty flawed strategy to me. Lots of improbable products turned into multi-billion dollar companies. Tiny videos set to music, watching other people play video games, liking pictures of food, and so on. It seems to me that it's less about the product and more about the execution. There are lots of Google products that could have taken off if they had been iterated instead of abandoned. Take Apple as one example. They develop a product vision, then spend several versions refining the thing till it goes from a curiosity to a must have. Or look at Facebook with Oculus, they had a vision and released headset after headset, till they found a feature set that had the user engagement metrics they were looking for. They didn't just give up once the Gear VR proved to be a dead-end, which is exactly what Google did with their VR efforts and then cleanly missed the boat Oculus is on now with the Quest.

Isn't that most private companies ?

Searching for sources of income ?

No, most private companies are not monopolies (although many would not mind I guess).

It seems online advertising has a strong network effect[1], where website owners use Google ads because that's where most advertisers are and advertisers use Google ads because it has most reach (and therefore tracking). This naturally creates a monopoly, just like with Facebook, twitter, LinkedIn, AirBnB, Uber etc.

[1] https://en.wikipedia.org/wiki/Network_effect

How exactly does a random piece of IoT junk that looks like it could come from any single one of countless fly-by-night manufacturers in Shenzen (namely the "Nest Hub" that's now running some version of Fuchsia, per the OP) turn into a money-making firehose for Google? Just curious, because I'm having trouble seeing how this could make sense.

Gee, how could the entry point to all the devices around you ever become a money making firehose?

The problem is that many good ideas only look good in hindsight.

In 15 years, we'll either say "Fuchsia was a brilliant move, I'm so glad my home runs on that OS" or we'll say "Remember Fuchsia? lol".

Here's what I mean. Acquisitions by Google in 2005: Reqwireless (mobile ads, merged with google analytics); Dodgeball (social networking, later Google Latitude); Akwan (Search tech); Skia (graphics library); Phatbits (Google Desktop); allPAY (???); bruNET (???); Android (cheap OS for mobile devices).

In 2005, which of these were going to change the world?

Advertising/influence depends on two things: 1) Knowing what you're thinking (e.g. the search request you speak to your Nest Hub) 2) Presenting messages to you (e.g. displaying search results/suggestions/ads on the Nest Hub screen in your home)

> Google develop 3+ different OS's?

It wont. From what I understand this new OS will replace others over time. Also it is way better than existing options.

That's a wild leap with zero evidence. Fuchsia runs on one unimportant production device and has basically zero usage. Saying it will replace all of Android and ChromeOS is...optimistic to say the least.

Project ‘Fuchsia’: Google Is Quietly Working on a Successor to Android https://www.bloomberg.com/news/articles/2018-07-19/google-te...

> For more than two years, a small and stealthy group of engineers within Google has been working on software that they hope will eventually replace Android.

https://news.ycombinator.com/item?id=19702968

> "And while Android is reaching EOL in the next 4-5 years, most of this work will be carried on in Fuchsia OS, which is set to eventually replace Android"

> "And while Android is reaching EOL in the next 4-5 years"

I wouldn't bet on it. Sounds like IBM OS2 all over again.

Check the context. The person I replied to is claiming there's "zero evidence" Fuchsia will replace other Google OSs which is obviously wrong.

May be. We have not seen future. But that is not the point and a person who is claiming there is "zero evidence" is mistaking absence of evidence to evidence of absence. Arsetechnica has published several articles on this topic in past and are available to read on Wikiepedia page for the OS.

General information suggests this is one of those long term things if successful would be massively game changing.

Better than all existing options? I'm skeptical that a kernel written in C++ is likely to be better than e.g. a formally verified microkernel such as seL4. If they ever rewrite it in Rust, then it might be more interesting.

Parts of it is already in Rust and formal verification could be done for this project as well. At that stage would you be ok with it?

If it improves the security of the device and means that Google will keep updating it, then that will be good. However, personally I have given up buying Google-branded devices. They only gave me security updates for my Pixel C for three years (2016-2019).

I am not a microkernel expert, but why is Zircon better than QNX as used in the final Blackberry handset?

QNX appears to already be in every market where Fuchsia wants to go.

Google could have bought Blackberry and used QNX in Android 5, rolling it up to Chromebooks. Why was this not the path they chose?

This seems like a good question, and I also hope someone with good knowledge/experience around this to give some good comparison.

I have given only very cursory look into QNX and Fuchsia/Zircon and it seems that one of the main difference is the capabilities-based architecture in Fuchsia. It's a big change in the API/conceptual model, but we'll have to see if that actually translates into a meaningful difference at the application level - e.g. whether it allows building new types of applications, or allows building meaningfully more secure/fast/reliable applications.

This would be another benefit of antitrust scrutiny for the tech giants.

They're not innovating at this point. They're twiddling thumbs.

If we broke up Apple, Google, Facebook, and Amazon, then they'd be forced to fight for survival.

Struggle births innovation.

Reply on news.ycombinator.com