How Google builds web frameworks
Filip Hracek

How big do you have to be to make this work? I’m guessing 25k devs might make this work by sheer virtue of throwing people at tasks. What about external code management, or does Google not have any? They must run multiple versions somewhere!

I don’t have and wouldn’t want 25,000 people to direct, but in my less than humble initial estimation, any organisation with 25k sw devs is too big not to fail, it’s a colossal number of bodies and I do wonder if the distribution of commits per-day varies greatly.

25000 * 20000 = 500,000,000 (half a billion) a year in cost, assuming most are in places where they can survive on 20k USD a year, which assumes most are on < 20k and only a few hundred or thousand top 100k. I’ve used USD for my estimates so to simplify the numbers, but I’d be surprised if Google’s software developer bill is less than this (it’d be good if it was < 5x this given their range.)

This is all without OPS, HR, marketing, sales, training, equipment, execs, etc. Basically it’s a lot of money to throw at problems that I’d imagine < 1% of businesses have access to. Aside from not being able to visualise how it works, I also question how applicable it is. Most businesses fail within 5 years and few even crack their first million in revenue.

I get it, Google makes considerably more moola than me, my business, most other businesses and probably a bit more revenue globally than quite a few nations. Maybe that is how they can undertake wonder-of-the-world software challenges, but maybe that also leads to lost focus.

I suppose I’m mostly shocked. It’s vastly different from my preferred style of work, what I’d advocate (branching development IMO is FTW). Maybe a link to a feed on Google revenue and profits would give the ultimate verdict. Maybe staff turnover stats would also be an idea as Ford also had good results, but their model introduced sustainability problems.