I think 18-36 months is probably a bit too large of a time span, but I think it’s a good idea (so much so that I did it for my own web app: http://irn.me/r).
I think the idea is most applicable when you have launched with a minimum feature set and have a list of things you know you are going to do. For example, having a mobile version of your site might make a lot of sense, but isn’t important enough to hold up launching.
I agree with you that announcing features beforehand is a bad idea—I’m slowly learning that you must resist that temptation if at all possible. However, if you know that nothing is going to stop you from implementing a feature, it might be a good idea to let your customers know that it’s on your mind (especially if it’s an industry-standard feature that will make them wonder WTF you were doing not to include it).
I think the idea is most applicable when you have launched with a minimum feature set and have a list of things you know you are going to do. For example, having a mobile version of your site might make a lot of sense, but isn’t important enough to hold up launching.
I agree with you that announcing features beforehand is a bad idea—I’m slowly learning that you must resist that temptation if at all possible. However, if you know that nothing is going to stop you from implementing a feature, it might be a good idea to let your customers know that it’s on your mind (especially if it’s an industry-standard feature that will make them wonder WTF you were doing not to include it).