XP is unfalsifiable, therefore it won't work

Christer Ericson wrote:

> The {Success, Failure} x {Full XP, Subset XP} table here
>
> http://www.hacknot.info/hacknot/action/showEntry?eid=53
>
> nicely summarizes the non-falsifiability of XP.

One form of pseudo-science is claiming the trappings of science when
debating a technical guideline.

Suppose I tell you to pour water in a beaker. I explain if you pour here, it
goes in, and if you pour there, it goes out.

Pouring water in a beaker is not a hypothesis, so you cannot claim, "Ah,
your theory that 'pour the water here it goes in the beaker' is not a real
hypothesis, because it only fails if I was 'doing it wrong', so it's not
falsifiable."

We suspect (as a measurement, not a hypothesis), that the more XP you do,
the lower the odds of burning up money for a long time, then failure. That's
primarily because the XP practices promote "fail safe", and "fail early". If
you can't get your highest business priority feature working, within a few
iterations, and under relentless testing, you don't have a project.
Otherwise, if you do, the odds that a project started that way will sustain
are higher. So the decision to try XP requires reason, not any appeal to
pseudo-science.

--
Phlip
[url]http://www.c2.com/cgi/wiki?ZLand[/url]