Britain’s Guardian newspaper analyzed a body of user comments on the articles published there, and the results were rather disappointing. In the first place, the number of commentators was very, very small compared to the number of subscribers. Then, it seems that a very large potion of the people who do comment are ‘habituals’. That is, they have time and leisure sufficient to write lots and lots of comments. Some wrote hundreds. So it seems that at least for the Guardian, most of the material it publishes remains uncommented on, and that a large slice of the comments are traceable to very few people. Why is this important here? It’s important because one of the elements that the altmetrics proponents have suggested for inclusion in new methods of estimating an article’s scientific influence is comments by the readership. Now the Guardian is a general interest newspaper, and the altmetrics people are concerned about scholarly materials, so beware apples and the other thing, maybe. But these results should make us aware of the need to be cautious.