[patio11]

Yes and no. Yes, he could say "I work for AmaGooFaceSoft on a new type of infrastructure product which most companies can't even conceive of but which makes a lot of sense when you have 10,000 engineers and several hundred thousand servers." Then the question is "Oooh ooh what kind of infrastructure product is that?" and the answer will be "Do you remember Map/Reduce or BigTable or Hiphop? It's spiritually similar to those but totally unlike any of them and if I tell you any more my boss will have my guts for garters. Really cool tech though. You'd love to hear about it... if you worked here. Of course, if I told you about it, you'd not understand half of the explanation, since it plugs into four other proprietary systems that you -- as a member of the general public -- will never know about."

[nostrademons]

What I do is list the general area of work as well as the skills used, so for example:

- Internal templating language. Details confidential. Skills used: LLVM, HTML parsing.

- Internal search-quality research project. Details confidential. Skills used: clustering, classification, unstructured data extraction, HTML parsing, AppEngine, Django.

- Internal prototyping framework. Details confidential. Skills used: webapp security, cross-domain web requests, rapid prototyping.

That gives anyone looking to employ me an idea of what I've been up to and what skills I bring to the table without giving away the keys to what my previous employer was doing. It may provide some tantalizing hints, but there's pretty much nothing useful there for a competitor to replicate it.

[staunch]

The whole thing is a joke. All of the big tech companies leak like sieves and it hurts nothing. MapReduce? BigTable? GFS? Hardly secrets at the time they were being worked on, not secret at all by the time they were in use, and completely public knowledge very shortly afterwards (not a "better part of a decade" like the comment above talks about). Same is true of virtually every project of note.

The vast majority of these projects wouldn't even help a competitor if you begged them to use it. Heck, a lot of these top secret skunkworks projects end up hurting the companies they're built for. They're unpolished, highly proprietary, and shoved down people's throats. The difference is often just that they're somewhat hidden causing people outside ascribe magic powers to them.

It's no wonder at all why companies foster secrecy. It makes everyone feel special and important. Good for morale and it costs nothing. Still a total load of bs 99% of the time.

[nostrademons]

I think secrets are like startups: the vast majority of them are worth nothing, but once in a while there will be one that's worth billions, and you usually can't predict which one that will be. So big companies try to keep everything secret just so they have this large portfolio of things they know that their competitors don't.

The vast majority of stuff I've worked on has been quite useless, but some of it has made millions of dollars, and it was very often the stuff that I thought was throwaway code or an interesting diversion that survives. Usually it's little details and not broad areas of work, though.