You argue against cherry-picking, and then you think you can prove something with one single example?

My dude, the other guy provided 100% more data than you!

There is a big difference between proof and falsification. The grant-parent wants to establish a general rule, whereas the parent only wants to falsify it. Usually, a single example is enough for falsification. -- There is only one caveat (that's why I said "usually"): The evidence used for falsification must be true, which in itself is a deduction from many detailed observations.

The real flaw in the argument of the parent is, that in a direct comparison between Singapure and Norway Norway still wins: If roughly the same recidivism rates are achieved in a very punitive and a very supportive system, the supportive system is superior according to utilitarian principles.

Surely you realise that the original point works equally well with any of the three given examples?

If the recidivism rate is 20% regardless of if the punishment is harsh or not, I don't see why the conclusion should be that the harshness is supremely important. The harsh punishment does not better than the lenient ones!

Likewise: with the rate is 70% and 20%, but the punishments are equally hard in both places, why on earth is the conclusion that clearly the punishment is universally effective at keeping the rate of recidivism low?

What the original posted was arguing is that harsh punishment is not an effective way of reforming criminals, as measured by the recidivism. The example merely needs to illustrate that recidivism and harshness don't go together.

What shakow did was simply adding another example proving this point, and then acting like they'd provided a contradiction.

>The real flaw in the argument of the parent is, that in a direct comparison between Singapure and Norway Norway still wins: If roughly the same recidivism rates are achieved in a very punitive and a very supportive system, the supportive system is superior according to utilitarian principles.

Wouldn't that depend on subjectively balancing costs and benefits of the other involved factors? For example, Singapore's system may (or may not, frankly I don't know and it's not important to the logical point I'm making) cost less than Norway's system. In that case, the question would be what value the society places on finding solutions that are supportive rather than punitive.

Right?

Perhaps. My argument contains an unspoken "ceteris paribus" clause, which is typically the case, when we compare things: that all relevant factors are on the table and everyone agrees on them. To generalise your argument: From a logical point of view, no conclusion is ultimately settled, because it cannot be ruled out that not someday someone comes up with a new and better argument.[1]

So when I pointed out a "flaw in the argument", I meant the argument as presented, without additional refinements (as you are trying to develop).

A minor detail: When discussing a normative question, it is important to carefully distinguish between normative and descriptive aspects. The descriptive aspects are everything that is the case. The normative aspects are everything that should be the case. Therefore, the question is what value a society should place on finding solutions that are supportive rather than punitive. -- And the idea that norms are up to a certain point "subjectively" is in itself a description whose normative relevance asks for an intense debate.

[1] There might exist a few exceptions, such as fundamental logical theorems. But this is too far off from our case.

> You argue against cherry-picking, and then you think you can prove something with one single example?

Yes, that's the whole point of a counter-example...

If I try to prove the all odd numbers are prime (look, 1, 3, 5, 7, ... – it works!), you telling me that 9 is not is enough to disprove my theory.

I'm sorry to say, but if you think that you can provide a counter example to the statement "there is no strong correlation" with one single additional data point, you've grievously misunderstood how correlations work.

>You argue against cherry-picking, and then you think you can prove something with one single example?

Well... he can! He can prove, with one example, that the other person's thesis doesn't fit the facts.

And he did that.

So, yeah.

Can you please explain your understanding of the word correlation? I'm very curios to hear how your reasoning for how statements about these can be proven by a single data point.

A single data point can be sufficient to contradict a thesis.

A correlation is an observation that seemingly unrelated events occur in proximity to each other, temporally or otherwise.

Some theses can be disproved by a single data point, not all, and I asked about a specific type of thesis.

I repeat: can you show how a single data point can contradict a statement about the absence of correlation originally claimed?

What? I feel like one of us has forgotten how we got here.

>The idea of punishment as a deterrent is hopefully dead and buried at this point.

That's an absence of correlation claim?

Yes: if punishment was an effective deterrent, then the natural expectation is that there is strong correlation relationship between the level of punishment used and the deterrence effect. Saying punishment doesn't provide deterrence only claims that punishment is not correlated with deterrence.

Maybe the examples shown gave the impression that an anti-correlation was claimed, but that is purely an invention of the reader.