37Signals recently wrote that once you get to a certain scale it gets nearly impossible to remove features and thus you should be really careful about adding new ones.

I disagree – knowing what features to deprecate is one of the toughest challenges in prod/eng/UX mgmt but good leaders have to make these types of calls. For example, at YouTube our user base is big enough that anything we do will be hated by at least some percentage of our audience. Remember, not even 100% of people like ice cream. Now of course you should respectful of the users – you exist because they support you – but ultimately you might piss some of them off, and this is okay.

So instead of fearing the removal of features, better to think about best practices for evolving your site in the face of a large user base. Some strategies:

1) Warning – give people advance notice when it impacts the way they use the site but sometimes you need to treat it like a bandaid and just pull quickly.

2) Be transparent – blog about why you’ve changed something and why you think it will ultimately be better. If you’ve got influential users or a community committee, talk with them first to try and gauge how best to communicate.

3) A little sugar makes the medicine go down – a feature removal without having added something cool around the same time makes it seem like you’re just taking, not giving. Try to deprecate features when you also have something cool to talk about.

4) Be data driven – sometimes you make mistakes and features need to return. But don’t let a few angry consumers dictate your product strategy. Look at the broader numbers – were you right or wrong?

5) Worst case scenario is to just hide it instead of removing – by sticking something on Account Settings or Advanced Menu, you can give the legacy users a place to go without increasing the number of people using the feature.