stark ( on November 7, 2007, 14:56:33 ) You are counting downloads of the two bibtex entries as "downloads" in the software overview. But the word "download" makes the user think the code for this software has been downloaded (and compiled/used) that many times, which is at least confusing.

sonne ( on November 7, 2007, 15:33:33 ) what do you suggest then? shall we just count everything as views?

stark ( on November 7, 2007, 19:35:45 ) "downloads" should be incremented only if the user downloads the source code. "views" should be incremented when a user watches the project details. I do not understand how a "download bib file" counter makes sense. Maybe in an area that is not publicly visible.

sonne ( on November 7, 2007, 21:24:11 ) I agree, a download bib-counter does not make sense. Views currently means click on the name (software details page). But we don't really have downloads ... Only some projects upload their .tar.gz - most of them only provide the project homepage link. So how about counting downloads and clicks on the project link as the same thing under "downloads" ? Any other ideas?

stark ( on November 8, 2007, 15:18:40 ) Posting a project to an open source software site without a link to the actual code is useless. Hence we have a link to the code for every project, wether this links to mloss.org or somewhere else. In any case it can be tracked. Personally, I think this counter business is useless anyway. It does not improve the program nor does it improve the site. I also doubt it improves a scientists street-cred (isn't this what rating is for?)

sonne ( on November 8, 2007, 16:35:02 ) Well there is a positive aspect about this: People love to see that their software is actually used and that people are interested in it. So it is kind of a reward system and may actually be motivating. But yes maybe we add a direct download url and require that either the archive or the url is entered. Then the counting will be: project page/bib/paper bib == view + 1 download link / archive click == download +1

sonne ( on November 11, 2007, 15:08:47 ) Is now implemented this way.

gtjemwa ( on November 29, 2007, 21:18:53 ) The tracking system for downloads is still confusing. In some cases clicking on the download url takes one to the project webpage. However, this does not necessarily result in an actual download. But the counter on mloss is incremented nevertheless.

sonne ( on November 30, 2007, 08:43:04 ) I am afraid there is not much we can do here. If the project does not have a direct download url (like sourceforge pages that require choosing a download mirror before download) I guess the best a project owner can do is to point to the download page. However, I think increasing the download counter is reasonable. After all we cannot know whether you don't press cancel while downloading or just remove the file after a download is completed. This is quite similar to someone going to the download page of a project and not clicking on the true download link.

Post a Reply