A Quora thread about how the big startups (Facebook, etc.) split out their developer time:

There is a part of Facebook culture that is actually against the “we’re only here to work on innovative, ground-breaking, glory-seeking projects” that grew out of a reaction against the perceived Google culture of the time (approx. 2006 – 2008), as well as a recognition that long-term, sustainable success is built far more on hard, thankless grinding than brilliant innovation.

As a result, Facebook actively respects people who do the “thankless, dirty work,” like rewriting spaghetti code, making build systems, crushing annoying bugs, etc (obviously, it ends up being much less thankless in that case). High-profile projects at Facebook get respect from the press and the outside world, but “doing the dirty work” gets you respect from your peers and management. At least while I was there, we had a tendency to prefer engineers for advancement who had spent considerable time “grinding it out” and supporting their peers.

Lastly, the occasional notable innovation or well-executed project is made possible because one’s peers have put in the time to ensure that all the mundane groundwork is done well (i.e so that critical subsystems do not fail on you when you are trying to launch something new and risky). This is appreciated on a peer-to-peer basis and, while I was there, most everyone took their turn doing “the dirty work.”