[Jack Simth]

Imperator Fyron said:
Jack:
Your percentages are unreal

Never said they were real.

Imperator Fyron said:
and don't properly reflect any sort of valid estimate

Never said they did.

Quit trying to imply a particular set of words to my behaf. It's annoying.

Imperator Fyron said:. A 99% effective system either means 99% of criminals are jailed (which says absolutely nothing about non-criminals, so isn't very interesting), or 99% of those jailed are criminals. It could also mean that 99% of innocent people accused are acquitted, but that probably isn't the intended meaning in context. It is a contortion of logic to assert that it means both that 99% of all criminals are jailed, and at the same time 99% of non-criminals are not jailed.

I stopped and defined my terms as I was using them. Going back and saying I used the wrong term is a void argument, as I chose the meaning to be used. Seriously, Fyron. You're assinging things to me I didn't say.

Seriously, though; if it were possible to go about ascertaining what portion of those who are found "no guilty" are *actually* guilty, or what portion fo those found "guilty" are *actually* not guilty, you could arrange to do so on an individual basis and get a 100% accuracy rating.

Imperator Fyron said:

The two are completely independent statistics that you can not validly roll into one called the overall accuracy and thus derive the rest of the assertions.

So you claim. I listed, quite blatantly, that what I was using were assumptions - which are pretty much by definition arbitrary. You're making claims about the real world. Can I get a source?

But then, if you actually read what I wrote, it doesn't apply to the math I used; I very clearly stated "You can play with the numbers some to get different results" - they vary; widely. There's a reason I pointed it out. Come on Fyron, if you're going to do this, do it right.

Switching from 99% on both sides (as you seem to think is silly) to a lop-sided set has a particular effect.

Imperator Fyron said:
From your 99% accuracy figure, it does not follow that 1% of the non-criminal population is jailed. At best, all you get is that 1% of the jailed population is non-criminal. At worst, you have absolutely no idea how many non-criminals are jailed, so you have no idea what percentage of those in jail are wrongly imprisoned.

The way I explicitly defined it, it actually does. When the defininition used does not apply, neither do the results of the math. So?

Read, Fyron. I'm reasonably sure you know how.

Imperator Fyron said:

You further compound the problem by asserting that 1% of all non-criminals are jailed; this is not a valid metric when assessing the accuracy of a legal system.

I didn't assert that they are; I said with the way I defined 99% accuracy, they were. There's a difference, Fyron, that you appear to be missing. As I defined my 99% accuracy, 1% of the non-crimal population does, indeed, end up getting punished.

Oh, and I didn't use the term "jailed". Quit putting words in my mouth; it's not polite.

Imperator Fyron said:
A more valid metric would be what percentage of innocent people accused of a crime are acquitted, and what percentage of people in both groups are actually brought to trial. Only then could you hope to come up with percentages of people in jail being innocent or guilty.

Again, when one declairs their definitions of assumptions and terms for a mental excersize (which I did), saying it's silly because of definitions that don't apply to a particular situation is alternately a "Well duh" or "So?" situation. It logically follows (in the actual sense, not the one most people use) that where the definitions used in a proof don't apply that neither do the results of the proof (if A -> B under C circumstances (which translates to A & C -> B under all circumstances), but C is false, the statement says nothing at all is said about B - this is basic logic, Fyron; you're nearly forcing me to spoon-feed it to you here). For example, the pythagorean theorm doesn't apply in a non-euclidean geometry (although there's patches for it, for certain ones) as you need a non-euclidean geometry to have a proper right triangle as defined for the proof.

But you've sufficiently demonstrated that you don't actual read the other side of the debate (or don't comprehend - if english is not your native language, the above doesn't apply, as it is purely a misunderstanding, and I apologize); there's no point in a debate. Have fun, I'm done.