Discrepancy

I am curious as to why there is a large discrepancy between the package user count (in the reputation) and the number of unique downloads in the package stats.

I waited until just after the system posted the daily results to the reputation to do my calculations.

Unique user count (reputation): 3263

Unique downloads (adding up all packages): 4637

Just so nobody thinks I am being petty... normally I would not care, but it just so happens that I am almost ready to advance again in the missions, needing 5,000 package downloads. Of course, it is using the Unique user count instead of the Unique download count, which will be the difference of moving up this year vs next year.

Dave

The difference is due to the fact that some users have downloaded more than one package of yours, but for unique users each one just counts as one.

Okay... I get it. Since it is applied to all authors, it is fair, however...

I think authors should get credit for unique downloads of each package. As it is now, an author that writes many packages is penalized if more than one package is of interest to a user and this penalty gets bigger with each interesting package they write. Eventually they could write enough interesting packages that they wind up getting no additional credit. I am quite certain that one of the goals of phpClasses is to encourage authors to write more interesting packages, however the current credit system does not support that.

Dave

Yes, it is an idea.

It would be a different metric that could be named total downloads, although the site just records the last download per package even if the same user downloads the same package multiple times.

Anyway, since it is a different metric, how do you suggest it being used on the site to encourage more authors to submit more packages?

To be clear, I am not suggesting credit for multiple downloads of the same package by a user. Each package currently has a count of unique users that have downloaded it and I am suggesting that is the count that should be used. Here is my thinking...

The current metric of unique users across all of an authors package has an emphasis on the importance of attracting new users, not on attracting new interesting packages.

A metric of the unique users per package changes the emphasis to the importance of attracting new interesting packages.

I can use daily access metrics on a site as a good analogy. An important metric is the unique daily visitors which counts each unique visitor once per day. Imagine if we instead put more importance on a different metric that counts each unique visitor only once in the lifetime of the site. The first one is a good gauge on the health of our site while the second one is only a measure on how well our site can attract new visitors. While it is important to attract new visitors, it is more important to keep them coming back.

You current system is a gauge on how well an author can attract new users of their packages, not on how interesting their packages are to the group as a whole.

The math is a bit more complex, however in simple terms it currently works something like this... Suppose 1 percent of users downloading an authors packages are return users. For each new package submitted, the effect is cumulative... so with the 1st package, all downloads are credited at 100%, the second is credited at 99% which continues until the 101st package which gets credit for 0% off all downloads.

I seriously doubt many authors would notice one way or another, I just happen to be one of those people that has to understand everything. So it really isn't a matter of encouraging more authors to submit more packages, it is a matter of not discouraging them. That said, I am not discouraged and I doubt any other authors would be either, I am just pointing out a flaw in the system.

Dave

Yes, I see what you mean.

As you also realized, it is hard to demonstrate that adding that new metric somehow could encourage more authors to submit more packages.

If you have any ideas about this, just let me know.