views vs. time

Very frequently, people ask how to "get more views", and one of the many answers given is to post regularly and parsimoniously. Does it work? Here's another analysis based on the same sample of 100 photos from 1000 people from yesterday, trying to address this question.

Each point in the graph represents one photo. The y axis is how many views that photo has (in log scale), and the x axis represents the time (in hours, also log scale) that elapsed between posting that photo, and posting the following photo in the same photostream. In other words, how long the photographer waited before posting a new photo, or also for how long the photo in question was "the most recent photo".

One can see again the split into "photos uploaded as a batch" (left side cloud) and "photos uploaded periodically" (right side cloud), corresponding to the two peaks in the red graph.

Clearly, posting parsimoniously does lead to more views per photo than posting in a batch. (But see the discussion in comments.) The distributions overlap, but it is quite evident that the right side cloud is higher than the left side cloud.

Using an arbitrary (but reasonable) cutoff of 1 hour, there are 20258 photos in the left cloud, averaging 59 views, while the right cloud has 11742 photos, averaging 199 views. (Ok, that's linear average, which is not quite the proper thing to do here. Using log averages one gets 12.5 vs. 69 average views, respectively, but these log averages are more strongly influenced by the view pseudocount I used. Bottom line is, now matter how you do the statistic, there is a clear difference between the two posting methods.)

Some technical details. Since many photos have zero views but I'm using log scales, I used a view pseudocount of 0.1. Quite clearly, many more batch-uploaded photos end up having zero views (you might need to visit the large size to see this). I also used a 0.1 second pseudocount for the time to the next photo, since many photos appear to be uploaded at the exact same second as the next one... Makes me wonder which uploading/editing tool does that.
Another technical detail is that since I used Excel to make the graph, I included here only 32000 points, which is roughly one third of the information in the sample. The way I sliced it means this graph shows information about one third of the people in the sample, not one third of the photos of all people.