Hacker News Reader

Hacker News

Comment by philip142au | Hacker News Reader

Parent Full thread

philip142au

Give an unsolved maths question. ;)

View on HN

Then the answer is: current math cannot solve this question.

Answer can be lots of things, it's easy to avoid actually answering the question.

Hey thanks for the comment.

If it can be done in 42 we will do it in 42, so NP is probably a hard one at this stage :)

The ultimate question is out of our humane reach unfo, but we'll try to cover the rest :)

You said "Any" question :(

Updated to "If it can be done in 42, it will be done in 42." on the website :)

So if it cannot be done in 42 minutes, it will not be done, right?

Since we qualify a question before proceeding, we'd politely decline if we can't do it in 42 minutes. Our hypothesis is that anything that takes longer qualifies as a consultation, and there are better platforms for people to find such white collar service providers.

What is your agency's policy for answering Newcomb's problem?

I just read about Newcomb's problem due to your comment. Maybe I'm being stupid but I don't think there's any paradox. Let's change it so that the first box contains only a tiiiiiny little speck of gold and nothing else, and you'll see.

For this demonstration, you get to play repeatedly.

Let's make you too smart for your own good. You start off with the strategy I think is stupid and wrong. So: each time you play, you select both boxes on the (stupid, wrong) theory that selecting both strictly dominates selecting just one. Of course, each time, there's nothing in the second box because your choice has been predicted.

After playing several hundred times and amassing a tiny little mound of gold flakes you think about whether you even know any shops where you can convert it, and you didn't even bring a zip lock bag with you, what are you going to do, go home with golddust in your pockets? You think about the hours of time that you've spent. Finally you wisen up and say fuck it, this time time I'll just select the one box.

Since your choice has been predicted, you finally get the million dollars, and you realize you were pretty dumb for wasting several hours on a few specks of gold.

You keep playing, selecting just the one box, and keep getting a million dollars each time. After you have more cash than you can drive home, you thank God for all your cash and go home.

This proves that the choice is simple and there is no paradox or contradiction. With my approach, do you see any paradox?

I share a similar view, that testing whether the examiner is lying is itself worthwhile. But while the analysis has several interpretations this question may pose the agency a small problem. They are answering a paid user and therefore have technically already opened box number 2. They cannot therefore lay claim to box 1, only dodge the question, lie, or accept their fate.

I'm sorry, I completely missed that you were relating it to the agency's problem.

I thought you were just giving any paradox as an example of a difficult problem, and could have given any.

I don't quite follow the analogy with this agency's policies, could you make it a bit more clear? What are the "boxes" in this case?

Not at all. It's correct that this problem is formally recognized as a false paradox. It is a thought experiment infamous for its illustrative misuse in relation to other areas. "What are the 'boxes' in this case?" - The unfortunate choice of fiat is one such area, and it is correct for practical purposes to discard the subject right here because answers will modify the service, or fail a test. So fail a test, it's not a problem. Ignore the unstated false equivalences between test criteria and aim to break the criterion that matters least. Different situations call for different choices, but standard consensus is to admit to 2-boxing. Different hybrid strategies are improved by personal values. By 1-boxing I don't value my time, I value the information. While I'm coming up to speed I might get a chance to use it. That could take a while though, I have a decent memory but I'm not a fast proofreader.

Interesting take on this whole scheme of things, thanks for taking the time. I'd add that we're qualifying the question before accepting payment i.e. if its out of scope, we politely decline.

On a side note, aren't we already doing the same when we go to white collar practitioners like doctors and lawyers for advice? I mean you pay the service irrespective of whether we like the advice or not. With FortyQ, we're giving you a way out of that with refunds.

Glad to hear that. The Freemium solution is certainly a simple one, but I didn't feel it was appropos. Personally I'd have to start writing a day ahead to have a chance at formalizing my question before the time limit. Expert advice does benefit from codified standards but it's hopefully dissimilar in that the question I brought features a double bind; It doesn't have a 'winning' answer. There are tons of regular answers, so an expert would just tell you not to fall for it. Education against them is one of the few ways I acknowledge of their constructive use, and I'd guess logicallee still won't be satisfied with my breakdown. I do what I can.

I see your point regarding a double bind, and it is something we see a lot in blogs I guess with conflicting viewpoints? Its true that there is a significant effort involved in formulating the question properly, but it also ensures getting a chiseled answer. Something we noted as well with paid questions: the amount of detail provided improves dramatically.

One thing that I have noticed working legal questions in the past weeks is that people more lost than they were when given all possible viewpoints and given the possibility to choose. More often than not they are looking for someone to tell them what to do. Makes sense though if you think about it.

Not sure I entirely got the 'Education against them.. ' point, could you elaborate?

Against the use of double binds. To me they are an indicator of hostile intent, although as demonstrated here and in the practice of law, adversarial interactions can be constructive. This abstraction is incredibly beautiful, but typical examples are not necessarily nice.

Do double binds automatically come in though, when you allow people of disparate backgrounds and experiences a platform to express opinion?

Could be constructive though as far as law is concerned.

What typical examples would you refer to?

They could, though I think much of that can be mitigated with effort. I believe the textbook example is domestic abuse, but I worry that the scope also coincides with fallout from poisoned well attacks and tribalistic recruitment efforts.

Thanks for your question, if you look at some of the examples on the site, our staple questions are contextual ones e.g. something the user is facing at the moment. That could be a legal, IT, finance problem or others that we can help with instantly, but I hope you get the idea.

For everything else, there is Wolfram Alpha :)

Reply on news.ycombinator.com