Video editors usage statistics: duty calls4 min read

WARNING: POLLS AND STATISTIC CAN BE BENT TO SAY ANYTHING
I can’t resist biting OpenShot’s bait. Jonathan, with all due respect, I think you haven’t paid enough attention in statistics class (or you paid too much attention and you are trolling). Reasons your conclusions may be wrong:

• In Ubuntu, PiTiVi is installed by default. Which means that your “users vs installed base” ratio is completely wrong if what you’re trying to measure is popularity.
  • Many users don’t do video editing (especially geekier ones), and I’ve met many of them in real life who didn’t even know that PiTiVi was installed on their computer!
  • Of course the OpenShot ratio is going to be higher, those who have actually gone through the trouble of installing it are among the target demographic who actually wants to do video editing!
• Debian’s demographics are quite different than Ubuntu’s.
• You are assuming that “more regular users per installs” == “the software is better”. It could mean anything, including (but not limited to):
  • “OpenShot users have more free time on their hands”
  • “It takes more time to accomplish something in OpenShot than in PiTiVi”
  • “2 + 2 = 5”
  • “OpenShot users do much more video editing”
  • “PiTiVi kills those who lack the necessary skills to use it, so only those strong enough can use it a second time”
  • “Users launch OpenShot often because it crashes”
  • …and various other conjectures of dubious value
• You are assuming that users who enable Popcon are a representative sample of the demographics you are trying to target. This is another confounding variable that weakens your analysis. This is like counting volunteers for a “scientific study on horse porn and interspecies erotica” and then extrapolating the figures to “the European population at large”.
• These numbers don’t take into account users who have multiple video editors installed (I have PiTiVi, OpenShot, KDEnlive, “Blender”, “Avidemux” installed,  for example. Each tool can accomplish a different thing).
• There is no reason to believe that ratios of users who enable popcon are the same for each video editor. For example, it could be possible that 100% of PiTiVi users enable Popcon, while 10% of OpenShot users enable it. In that theoretical scenario, OpenShot’s userbase would be widely underestimated.
• We could argue that being installed by default on a major distro qualifies as some sort of unquantifiable indicator of quality. As such, I could naïvely say that the bare fact that PiTiVi was chosen as the default video editor on Ubuntu means it has been “judged” as “better”. But making such a debatable claim would be preposterous (and I would be accused of bribing senators and eating babies).

What about the following assertion? :

“I believe this data supports what other polls have already suggested: OpenShot is the best Linux video editor available today.”

I can play with pretty graphs, too. Everybody stand back. I can use Gnumeric.
Here is the graph you base yourself on:

Here is the same graph with only the absolute number of regular users:

Guess what? “More regular PiTiVi users than OpenShot and KDEnlive, combined“.
But wait, remember I said that Popcon is a weird metric? Consider this: PiTiVi, Simple Scan and gedit are all installed by default on ubuntu desktops. Yet, the number of installs varies! PiTiVi and Simple Scan both have about 275 000 installs each, yet gedit’s number of installs goes over 1.5 million!

Let’s (ab)use even more statistics!

I’m just playing around with numbers to prove my point: all those metrics are bunk and cannot seriously measure the quality of software. Do not believe anything I say in this section.

Lines of code

PiTiVi has 25 thousand lines of Python code (22% of which are comments) versus OpenShot’s 8 thousand (33% of which are comments)! Obviously that means our application is better! No wait, does it mean that our application requires more lines of code to do similar things? Damn!

Bug count

Surely, the amount of bugs an application has is a good indicator of quality? (No, it’s not).
Let’s see, for good measure*, we want to count bugs in both upstream and downstream (Ubuntu’s) bug trackers, and we want to separate “crappy” bug reports and “wishlist” items from “normal” bug reports. If we didn’t, the bug count would look like this:

When you clean up the numbers a little, you end up with 160 “real” PiTiVi bugs and 114 “real” OpenShot bugs. OpenShot wins:

*: not a good measure. I just picked today’s bug count, which doesn’t mean anything anyway. And even if it did, it doesn’t. Also, you need special privileges (I do) on Launchpad to see “private” bug reports, so my numbers probably won’t match yours.

Number of code monkeys

How many people with more than 10 commits? OpenShot = 5, PiTiVi = 18. Oh wait, that statistic is bunk too: it can include old committers and translators.

Let Python decide who’s the winner!

Python 2.6.6 (r266:84292, Sep 15 2010, 15:52:39)
[GCC 4.4.5] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> pitivi = len("pitivi")
>>> openshot = len("openshot")
>>> if openshot > pitivi: print "OpenShot is greater than PiTiVi"
...
OpenShot is greater than PiTiVi
Let’s call it a draw, shall we?