As pointed out by RetractionWatch, AltMetrics even tracks the metrics of a retraction notices.
This retraction notice has an AltMetric of 9 as I write, and it will grow with every mention on blogs (such as this) and Twitter. Even worse, even just one blog post and one tweet by Retraction watch was enough to put the retraction notice “In the top 25% of all research outputs”.
In my opinion, this shows how unreliable these altmetrics are. They are based on the false assumption that Twitter and blogs would be central (or at least representative) of academic importance and attention. But given the very low usage rates of these media by academics, this does not appear to work well, except for a few high-shot papers.
Existing citation indexes, with all their drawbacks, may still be more useful.