News for hackers. Stuff that matters.
Re:The Subscription App Paradox by josephg on Sunday September 03, 2017 @07:26AM (#15160878)
Yes, this is a constant sourse of amazement for me too! The chrome team at google is well over 1000 people. (The opensource code base now has over 800 committers). The initial release had 100 committers[1], so probably about as many man hours have gone into google chrome in the last year than were put into the project in its first ~5 years of existence or something. In the last year just shy of 1.5 million non-blank lines of code[2] have been added to the browser, which is more code than was in the whole browser when it was first launched, webkit included.

But what does all that new code do?? I'd challenge anyone who's not a developer to name a single new feature added to chrome in the last year. I'm a web developer and I struggle to name more than 5 changes.

Its easy to pick on chrome because the numbers are mostly-public. But the same is probably true of facebook.com, ms office and all sorts of other big company products. As features compound, the marginal value of each additional line of code added becomes vanishingly small. (All the really useful features have already been added, and project iteration gets much slower in big projects).

There are thousands of engineers working on facebook.com. If they all quietly left and were never replaced, how long do you think it would take before anyone in the public would notice? 6 months? 1 year? Longer?

And if thats the case, what a waste of that huge pool of talent. If you could leave your job without any of your users noticing or caring, what a waste of your education and your potential. Quit and start that company you've always been dreaming about starting. Go do literally anything of substance.

[1] https://www.openhub.net/p/chrome/contributors/summary [2] https://www.openhub.net/p/chrome/analyses/latest/languages_s...

Re:The Subscription App Paradox by icebraining on Sunday September 03, 2017 @12:48PM (#15162436)
Just taking a cursory look to their blog, there's PWA/WebAPK, Payment Request API on desktop, Web Share API, WebUSB, WebBluetooth (in progress), improvements to WebVR, Network Information API, new headers like Clear-Site-Data, the new Web Push Encryption format, out-of-process iframes, headless support, native notifications on macOS and IndexedDB 2.0.

And of course, there are countless bug fixes and other small improvements, and all of this has to work on six operating systems.

Re:The Subscription App Paradox by josephg on Tuesday September 05, 2017 @11:07PM (#15180907)
To be clear, I'm not saying Chrome's thousands of engineers aren't busy. They're obviously adding literally millions of lines of code that does something. I'm saying the features they're adding are very marginal. All the important stuff was added years ago. 1000 world-class software engineers can help solve some of the world's really important problems, and writing a "web share API" or bringing back web notifications sounds like a tragic waste of their talent.

"I saw the best minds of my generation consumed putting advertising next to pictures of cats"

Re:The Subscription App Paradox by dannyw on Sunday September 03, 2017 @07:49AM (#15160962)
The incremental value of one software engineer working on tiny things can still be positive if they're working on a product that's used by hundreds of millions of people.

For example, a small feature or redesign that increases engagement of a feature by 10% can make millions for Facebook.

Re:The Subscription App Paradox by lugg on Sunday September 03, 2017 @08:42AM (#15161142)
I think you could argue that small things like that can be done by a small team.
Re:The Subscription App Paradox by jooke on Sunday September 03, 2017 @10:01AM (#15161439)
These small optimisations are pretty independent though. If you double the number of engineers you can do twice as many of them.
Re:The Subscription App Paradox by dangoor on Sunday September 03, 2017 @09:44AM (#15161373)
Does any of that Chrome code include Chrome OS stuff? Google has been putting a fair bit of work into that.

The web platform has been growing (progressive web apps, more features in JavaScript include WebAssembly), and yes I'm sure there are some new features in Chrome. Does that include the mobile app codebases? Chrome for mobile has been adding features.

1000 people is a lot of money, so Google clearly sees value in what they're doing.

Re:The Subscription App Paradox by josephg on Sunday September 03, 2017 @10:18AM (#15161531)
I suspect it doesn't include ChromeOS but does include the chrome android app. Someone please correct me if I'm wrong.

They've definitely been busy, somehow. And they've definitely added a few marginal features here and there in the last year (position:sticky, grids, css modules, headless snapshot rendering all come to mind for me). But 1.5 million code reviewed lines of programmer effort seems disproportionally expensive for what they've done.

How many startups could you start given 1000 google engineer-years of effort? 100? Don't get me wrong, I like webassembly. But the output of 100 new startups with world class engineering teams and 1 year of runway sounds much more interesting by every metric that I can think of.

Re:The Subscription App Paradox by dangoor on Sunday September 03, 2017 @11:16AM (#15161838)
How many startups can take advantage of improvements to the web platform to ship to more devices or ship faster?

How many more "uber for cats" startups do we need? (That's purposefully not a fair characterization… A lot of startups are doing less interesting/useful work than improving the platform.)

A team of 10 people can creatively put together some really cool stuff. But you know what? Not all problems can be solved by a team of 10 people. And what happens for the ones that manage to bring in either revenue or more VC? They start hiring. And then you can complain about "why does company X need 1,000 people to do what they do?"

I do get what you're saying, but I'm pretty sure the list of things those 1,000 developers produced and those 1.5 million lines of code represented would be substantial if you dug in. A great deal of it might not have any appeal to you, but undoubtedly has appeal to someone.