Sturgeons Law Disproved

[Note to WikiGnomes: the usage of 'crud' below is actually the correct one. SturgeonsLaw is often misquoted as using the word 'crap' instead of 'crud': http://www.jargon.net/jargonfile/s/SturgeonsLaw.html]

SturgeonsLaw leads to a contradiction, and is thus false:

Assume that NinetyPercent? of everything is crud.
Gather up all of the stuff that you can find, and separate it into the 90% which is crud, and the 10% which is not crud.
Now, examine the 10% which is not crud:
- According to SturgeonsLaw, NinetyPercent? of this remaining 10% is crud. However, we have established that this particular sample is crud-free. Hence, we have a contradiction.
A similar argument applies for the 90% of the stuff which is previously identified as crud.

QED.

This is the fallacy of subpopulations, and is equivalent to saying "Statistics show that 90% of the population speaks Swahili. Now examine the 10% that do not speak Swahili. By the Swahili Law, 90% of that population should speak Swahili. But we've already established that they don't. Thus, contradiction."

Is this a workable example? In the Swahili example "the population" can reasonably be assumed to apply to one particular set, where in SturgeonsLaw "everything" is sufficiently vague to cause the recursion.

In other words, all that this demonstrates is that, to be accurate, any law of this sort can't be a recursive law, that's all. It applies to some universe of discourse (assuming one can be defined), not to subsets thereof.

Well, what if SturgeonsLaw said 100% of everything is crud? It would recurse fine, though then it would be reduced to some sort of set-equivalence theorem of cruddiness. And don't forget that a subset can be the same as the original set. -- francis If so, would that make SturgeonsLaw crud or not? Sure, why not? But just because something is crud doesn't necessarily imply that it's false.

Awww, you're no fun, Doug!

:-)

[More simply: 90% of everything is crud. 10% of everything is not everything.]

True statistical representations include an additional dimension: the confidence factor. Recasting SturgeonsLaw with true statistics would be something like NinetyPercent? of everything is crud with NinetyPercent? confidence That means that 90% of the time that you take a random sample of anything, 90% of it will be crud. This allows for recursively applying the rule to subsets without a problem, so long as the subset is a random sample. ;)

It is surprising how often exactly this sort of fallacious reasoning turns up in 'real work', and how often it is accepted. People are not, in general, very logical.

[Someone is confusing "everything" with "anything". If Sturgeon had said that 10% of anything was crud, then the law could be applied to the 10% that wasn't crud. But he didn't so it can't.]

A couple things may save SturgeonsLaw:

If we redefine it to only cover statistically valid samples; then it need not apply to datasets that are already segregated by cruddiness.
It may be metaphysically impossible to segregate crud from non-crud; for any sample we could choose, there exists a universe in which ninety percent of it is crud.
- Alternatively, the act of segregation violates Heisenberg, and will cause 90% of the non-crud sample to spontaneously turn into crud (and likewise, 10% of the crud sample will become non-crud). We don't need to invoke non-applicable quantum mechanics concepts for this to work. All we need to do is reformulate SturgeonsLaw to something closer to the ParetoPrinciple: "In any naturally-occurring sample, 90% of the items are significantly more cruddy than the rest.") [Of course, that's a tautology: "90% of the sample is below the ninetieth percentile in some measurable parameter".] Not really. The supposition is that there is a discontinuity. It might be more clear to invert it: 10% of the sample comprise, say, 90% of the total quality of the set; 90% of the sample comprise the remaining 10%, and are therefore, crud. [Unless you live in LakeWobegon?, where all the children are above average...] :-)
Crud may be in the eye of the beholder; for any sample we could choose there will be an observer who claims ninety percent of it is crud.
Cruddiness may be sample-relative (in which case SturgeonsLaw is a tautology; stating that for any given sample, 10% of it will be better than the other 90%).
As we approach the speed of light, the amount of crud in any given sample will approach unity.
at least 90% of everything is crud, because everything is crud. Though in this case the 90% figure is rather arbitrary.

What about saying that if you take the 10% which is not crud, you can't apply SturgeonsLaw to that, because that isn't everything anymore?

[That would fall under the statistically valid sample clause.]

Regardless, even crud can be usable. I'm typing this on Windows after all. . .

Damn! My crudometer broke. Made of cruddy parts I bet.

There is a 90% chance that the main argument of this page is crud.

This is just Zeno's Paradox all over again. Then again, perhaps my "incrudulity" has been compromised. Hah hah.

When after the first iteration you take the 10% that you declared "not crud", and consider them a whole, your definition of "crud" changes so that the law still holds.

Does this mean that 90% of Sturgeon's Law is itself crud?

yes Sturgeon's L is crud

So, aw isn't crud?

No, but AWT (JavaAwt) is crud. Use JavaSwt instead.

CategoryWhimsy