While small conferences don't have the manpower to check for registered participants, the vast majority of participants do register. But as conference costs go up (for reasons like extra proceedings sales going down dramatically) and grants getting smaller and harder to get, there has been a mild increase in people skipping out on registration. If this trend continues, the problem feeds back onto itself, as those who do pay feel foolish, and we have a serious concern on our hands.
If you don't get a copy of the proceedings or eat at the conference meals you might think that you are not costing the conference anything by attending and so don't feel guilty by not paying. But conferences have some fixed expenses and most of the other expenses are cheaper per participant if there are more participants so you are costing your fellow researchers real dollars by not paying your fair share.
Sometime the registration fee can make the different in a decision on whether to attend a conference. If so talk to the conference organizers; if you explain the situation sometimes arrangements can be made. Better to attend at a reduced rate than not attend at all.
On a side note, I'm off the web next week. Microsoft postdoc Ryan O'Donnell will guest blog. Enjoy.