Here's what i find annoying. There are so many StackExchange sites now. I have been participating in SO for about 2 years (15K). One tag i like to watch/answer is "machine-learning." ML questions are now spread across so many SE sites: Cross-Validated (Stats), Machine Learning, Computer Science, Computational Science", Numerical Modeling & Simulation--and more on the way, Information Retrieval, Bioinformatics, Computational Linguistics*. Because of this, SO is a lot less interesting for me. It's time consuming to track down the Questions from across all relevant Sites; about half of the questions have comments telling the OP to post elsewhere ("maybe this belongs on ....") and/or multiple votes to close.
And the worst part--as a programmer, i would much rather have the computational linguist answering ML questions on SO than on a specialist site (of which there is one). Maybe those guys benefit by becoming better programmers, but i certainly benefit from their deep knowledge of how to solve ML problems. Their answer is code, just like mine, but from an entirely different perspective, more often than not it's humbling.
It is possible to set up a cross-site question filter on stackexchange.com (http://stackexchange.com/filters). As far as I can tell, it only allows filtering by tags, not full text, so your mileage may vary.
And the worst part--as a programmer, i would much rather have the computational linguist answering ML questions on SO than on a specialist site (of which there is one). Maybe those guys benefit by becoming better programmers, but i certainly benefit from their deep knowledge of how to solve ML problems. Their answer is code, just like mine, but from an entirely different perspective, more often than not it's humbling.