Measuring the scientific-software community [LWN.net]

Measuring the scientific-software community

Posted May 20, 2016 16:48 UTC (Fri) by ScottMinster (subscriber, #67541)
Parent article: Measuring the scientific-software community

So the installed base corresponds to the area under the spike of the download curve, but above the post-spike baseline level.

I'm not sure I understand this correctly. Who are the people downloading the software at the "baseline" level if not users? If anything, my intuition says to throw out the spike as experimenters and the curious, but not necessarily your core users. The curious might become users, but they are also likely to not. It's been my experience that people who use and depend on a piece of software resist updating because they cannot afford for the upgrade to break something.

to post comments

Measuring the scientific-software community

Posted May 20, 2016 17:12 UTC (Fri) by sfeam (subscriber, #2841) [Link] (1 responses)

What causes the baseline download rate? Indeed a puzzling question. I maintain several scientific software packages, ranging from reasonably popular to truly obscure. In all cases there is a continuous level of downloads. Who on earth would periodically download an obscure package that has not been updated in years? I have never been able to figure that out. But in any case the description of a baseline level with spikes following each new release matches what I have seen myself.

Measuring the scientific-software community

Posted May 26, 2016 5:44 UTC (Thu) by dlang (guest, #313) [Link]

people don't all discover the software at the same time.

You also have people who start using it installing it on more systems (and yes, many people do install from the Internet instead of from a local copy)

Measuring the scientific-software community

Posted May 26, 2016 14:02 UTC (Thu) by oldtomas (guest, #72579) [Link]

> I'm not sure I understand this correctly. Who are the people downloading the software at the "baseline" level if not users?

This mystified me at first sight too.

But actually, the "steady trickle" are those downloading the thing, saying "oh, cool" and *perhaps* keeping using it or perhaps going on to other things.

You usually don't re-download the thing when the version hasn't changed (more so when some sort of cache, e.g. a decentralized VCS or your browser's web cache are in-between).

But if you are actively using it, you sure will download the next version, so yes, the interesting information is hidden somewhere between the peaks and the baseline.