I agree with Brian's comment that it can be worth predicting small quakes. But, in my opinion, it is hard to impossible to get a feel just by looking at whether or not a prediction for a small quake is significant. A record of many predictions pretty much has to be statistically evaluated....here, by Roger. If someone does a prediction for a large quake, and a large quake occurs, then a single prediction can stand out as "interesting"...Tim's and Petra's separate predictions related to the Parkfield quake stand out, as does the first of Shan's Great quake predictions..
Chris