Comment by crucini | Hacker News Reader

Comment by crucini | Hacker News Reader

This article is a fine example of IYI - "intellectual yet idiot" as Taleb terms it. Verbal cleverness + plausible statistics + apples/oranges comparison = what? Equals: "We have the answer to everything, but sadly the rulers and electorate aren't as wise/clever/compassionate as us." Others in this thread have pointed out specific fallacies in this article - for example there are prisoners in the US using chainsaws and axes; there are prisoners in Norway in solitary confinement who would be in general population in a US prison.

But let's zoom up to the bigger syndrome. Notice the author quotes at least one offender, but doesn't bother talking to any corrections officers. Did it occur to him that someone who worked in a prison for 20 years might know a little bit more about corrections than someone who read a bunch of studies and statistics?

Symptomatic of a broader problem - the chattering classes, who consume and generate information, are increasingly cut off from the real world, and increasingly influential. Of course it's easy to have opinions about how something "should" work when you have no experience and no skin in the game.

20 years of individual experience does not give you nearly as much data as "a bunch of studies and statistics". Being glib about the scientific method doesn't make you better than it.

Adam just raised seed funding to build the next great mobile app. He needs a high performance back end which will support rapid iteration. He looks at two candidates for tech lead:

Bob has 20 years experience in server software development and led the back-end development at a previous mobile startup.

Charlie is an econ PHD from Yale and has many statistics at his fingertips about software development. He has never written any software.

Who should Adam hire?

These scenarios are not analogous.

I think there is something deeply funny about reading a discussion on how someone else is an "Intellectual Yet Idiot" and then watching them immediately turn it around with a bizarre seed-funding analogy (you might even say an apples/oranges comparison!)

Those with 20 years of experience have "a bunch of studies and statistics" too. They are better equipped to interpret the data.

interpreting data is not part of the mandatory skillset of LEOs

Trying to build programs without real-world experience causes problems. And real-world experience isn't part of the mandatory skillset of those who interpret data.

Let me give you a clear, humorous example of something real-world experience will give you that studies will not:

http://www.bmj.com/content/327/7429/1459.long

Real world experience is not the thing that gets you to put on a parachute when jumping out of a plane. Parachutes weren't invented by people jumping out of planes with different things strapped to their backs until they found something that worked. They built thing and changed them and then tested those changes by building/testing wind tunnels. My god the amount of effort it must have taken to develop a way of packing a parachute so that it would open reliably must have been enormous.

On the other hand anyone with nothing but real world experience jumping out of airplanes with parachutes on would say, from experience, that parachutes are 100% safe. Someone could easily come along and develop a much safer parachute, and get zero adoption from the people with "real world experience" who assume that their experience is unbiased and sufficiently powerful to be representative.

The quantitative model is the only force by which humans reason. Anything that you can't model is merely happening to you, and that's what you hold in high esteem?

So what you are saying is that human experience is worthless? Is that really a position you want to commit to? Or are you willing to admit that practical experience has some value?

I was once framed for theft by a co-worker. (The investigation exonerated me.)

We all have skin in the game. The justice system is frighteningly imperfect.

That's not exactly what I meant by "skin in the game". What I'd like to offer to any would be reformer and rehabilitator:

1. You will pick an inmate who you think has maximum chance of reform. He must be someone facing 5 years or more imprisonment. He must consent to the experiment.

2. He will be released into your custody and live in your apartment/house.

3. You will receive a check for the estimated total savings you created by housing this inmate. Typically 6 figures.

4. You are responsible for his good behavior for his lifetime. If he commits any offense, you will pay the same penalty as the offender. In addition you will be financially liable for restitution if any, and the costs of imprisoning him.

You have divided society into two: the criminal and the innocent. I do not detect an iota of empathy for the criminal in your analysis.

Our imperfect justice system often allocates individuals into "criminal" and "innocent" buckets unfairly. Don't assume that you and those you love are immune.

Prisoners have rights. Prisoners should be treated humanely. Prisoners are human.

I don't quite see how your comment relates to mine.

I just offered you a scenario where you can exercise compassion and make a lot of money at the same time. But you would have to shoulder responsibility. Would you do it?

What if you were that prisoner?

An impossible scenario? You could never be a criminal, so measuring the outcome need not take into account the prisoner's stake?

If I were the inmate, I would most likely accept "early release" to a would-be reformer. I would happily sleep on the guy's floor to get out of even a minimum security prison.

I see a number of advantages for me. Not least, I have someone who really, really wants me to succeed on the outside.

But you haven't yet told me what you'd do, either as inmate or prospective reformer.

It's not particularly hard to find people serving 5+ years in US prisons who one could gamble are no more or less a threat to society that the average person.

Having said that, I could make you the same offer, except I choose for you a as-of-now non-offending police officer, corrections officer, doctor, fireman, military service man or other upstanding citizen, for whom who will take life-long responsibility. If they commit a crime your penalty will match theirs.

If that doesn't sound super appealing, it might have something to do with playing Russian roulette, rather than an argument against rehabilitation.

I think both your points are valid: the "Black Swan" nature of the criminal liability, and the baseline risk. How can we reformulate the scenario to avoid both? Also, I want to emphasize the reformer's level of belief in his reform, rather than ability to pick the perfect inmate.

How about these changes?

* No criminal liability for the reformer - it all gets converted to civil financial liability. So a murder might be $1M; simple theft might be triple the value of the stolen article.

* Reformer's initial payment includes an offset for "average criminality", using the above dollar conversions. So if the reformee exhibits average criminality in the remainder of his life, the reformer only pays back that offset.

* Inmate is picked by lottery, excluding those with strong signs of unreformability.

I'm afraid these mods make the example less intuitive.

Reply on news.ycombinator.com