Joel Spolsky on Bugtracking and Linus Kicks Ass

November 7, 2002

@ 11:58 PM

Tracking Developer Bug Metrics

It is (or should be) a common adage of scientific and social research that the act of measuring a phenomenon can significantly affect the outcome of the event being measured. This is especially true for any metrics related to job performance. If testers are rated on how many bugs they open then they file a bug for every typo or open several bugs for slight variations of the same bug. Support people rated on how many newsgroup/mailing list questions they answer end up answering several easy ones and skipping the hard ones because they take longer to answer. The same goes for call center operators, programmers, etc.

However Joel's response to this is seems to be the typical "throw the baby out with the bathwater approach" which is a common human reaction. The same mentality that suggests abolishing patents as the right solution to frivolous patents or abolishing copyright because the RIAA companies sell CDs at too high a cost is the same one that discounts a metric as useful as which developers are generating the most bugs per LOC.

Of course, Joel is creating a commercial product and his choice was a business decision. From a business perspective I agree that the risk that people would misuse such functionality and thus end up giving his product a bad reputation amongst its users was high enough for him to not ship that feature. However, this doesn't mean that tracking bugs per developer in addition to other metrics such as developer output (e.g. if Kevin has 20 bugs a month vs. Dan's 5 bugs but Kevin writes 5000 lines of code a month to Dan's 400 then Kevin is in better shape than Dan even with his higher bug count) is not a worthwhile activity that could benefit the company in the long run.

#

A Trip Back In Time

The online NES emulator took me back in time as I played Contra, Super Mario 3 and Kung Fu from my web browser. This is probably the first Java applet I've seen that doesn't suck. Kudos to Norbert Kehrer

#

Linus's Flaming Fist

It's always interested me how differently leadership affects areas of human endeavor. From countries to corporations the quality of the person at the top is typically a deciding factor as to how successful these entities are. In the Free Software world I always look at RMS/Hurd and Linus/Linux as a fundamental example of what a difference in leadership can mean.

Emails like this are why I wish Linus was my boss or a co-worker.

#

Idiotic Slashdot Story of the Week

Microsoft takes on PDF: I don't know which was dumber the "article" and "analysis" or the Slashdot responses. Even after raising my threshold to +4 I was still reading about three idiotic posts for every intelligent one. I'm particularly impressed by the fact that no one put two and two together by reading and comparing a previous Slashdot article with this one.

Actually, the story isn't as amusing as the fact that Slashdot sat on the J2EE vs. .NET benchmark story until they could find a counter article.

#

« Trip to Atlanta and More "Daddy Dearest"... | Home | C# Team Announces New Language Features ... »

Comments are closed.

Dare Obasanjo's weblog

"You can buy cars but you can't buy respect in the hood" - Curtis Jackson

Navigation for Joel Spolsky on Bugtracking and Linus Kicks Ass - Dare Obasanjo's weblog