Hacker News Reader

Comment by EGreg | Hacker News Reader

According to Bayes' theorem, we can't really know!

There is strong evidence in one side: it's highly improbable to win twice.

Yet there is strong evidence on the other side: it's very very hard to cheat.

However, we may never know. The same is true of any controversy - including religion vs atheism - where there are really strong arguments on both sides which each side drops like bombshells: "it's very improbable that life began from non replicating matter by random chance" and so on.

Downvoters: My comment answers the question posed by the OP, and invokes interesting mathematical concepts and also shows that this issue is far more general. WHY do you knee jerk downvote without even responding? Four downvotes in 30 seconds. This is on track to become the most downvoted comment of all time, with no explanation :)

View on HN

I was one of those downvoters because your comment adds nothing but pretends to be very insightful. You aren't actually computing likelihoods and multiplying them together, you're just pretending to. Bayes' theorem says we really can know: you just need to see enough suspiciously successful winnings to be able to conclude that something was happening outside of the standard lottery process. Bringing in religion to this entirely unrelated field is pointless.

"We can't really know" is a ridiculous thing to say about lottery cheating, when often a small investigation will show that yes, someone is definitely cheating. It's practically anti-scientific to phrase it as a shrug of the impossibility of knowing.

It might be safer to say that it's hard to know how much undiscovered cheating there if there is no discovered cheating.

The religion example is even less helpful, as religion isn't evidence-based.

Suppose the lottery system is designed in such a way that it really IS practically impossible to cheat.

The numbers are pulled out of a machine and broadcast live and there are tons of witnesses. Or maybe the numbers are generated as a repeatable function of some publicly known events. And so on.

Suppose the chances of winning this lottery fairly are one in 100,000,000.

Then someone wins 3 times out of 20 tickets.

What would you say?

People try to find the flaw and they can't.

What if you DIDNT know the lottery was designed to be perfect? Are you sure you wouldn't constantly suspect the WRONG choice?

The system that you describe in no way resembles the lottery system as it exists today: it's so far from reality that I'm not quite sure why you're even bringing up this fantasy representation of a lottery.

The real lottery abuse - particularly scratch-offs, which are what is being abused here - is not a match-up between "high probability of not winning" and "unable to cheat". It is between "high probability of not winning" and "multiple fairly obvious avenues for cheating". Among them are the two mentioned prominently in the article: Micro-scratching and taking winning tickets and telling customers that they were losers, then claiming it for yourself. Both of which are avenues easily available to the clerks who have suddenly found themselves to be in this "miraculous" state of repeated wins at very low participation rates.

Why are you bringing up some fantasy straw man version of the lottery to defend it?

Because this is HN and people here are interested in exactly such generalizations.

It's mildly interesting to consider the ways in which scratch-offs can be gamed. One can collude with a shop owner to peer behind the scratch-off material with a machine, for instance. It is about as interesting as figuring out how a biometric ID can be gamed by a simple replay attack, storing all the aspects of your fingerprint and replaying it later.

What's MORE interesting is if the lottery was far far harder to game. The general principle is fascinating. We think we can KNOW the truth but using simple math we can SHOW that when evidence on both sides piles up and yet contradicts the other, you cannot recover and somehow come to KNOW the truth.

Thus you see the situation today when people confidently assert one side or the other, but each side has really strong arguments for its side.

Each side will say the other's arguments are bunk, but rarely actually engages with them in depth.

I still don't understand.

Winners can be investigated where it may be found out that they colluded with someone in the lottery bureau, used a pin and magnifying glass to peek in scratch offs at the 7/11 they work at, etc. The article enumerates these.

If they didn't cheat at all or just cheated very well, then of course nothing can be determined. Just like any other crime. But that's not very insightful. There are unsolved murder mysteries.

Forget scratch offs. What if I told you that the lottery system was actually not breached. Every time you look for a flaw, the system has all indications of not being susceptible. Understand?

And yet someone won 3 out of 20 times and it multiples to a staggeringly unbelievable probability.

You will keep suspecting that you were wrong, long after you make a conclusion. Do you understand? The stronger the evidence on both sides, the less ability you have to remove doubt, EVER!!!

Only in the absolute sense. You cannot provide canonically that an unobserved event happened or didn't happen.

That's why we have an investigative process. The investigator or auditor looks at patterns and notices that a guy wins the lottery multiple times. Then you dig and notice that the guy buys or claims tickets at a single or small collection of retailers. Then you dig and find out that there are personal relationships between the "winner" and the people in the store.

At that point, there are options:

- The lottery can amend it's regulations and review how the games work and make appropriate changes.

- The lottery can investigate the retailer and take action against the retailer's license to sell tickets. Typically the lottery has broad discretion to take action there with a legal standard that has a low bar.

- The lottery can pursue a civil action against the retailer or "winner". Civil actions require a jury to meet a preponderance of evidence standard.

- The police or IG can pursue a criminal case, with a high standard.

I don't think that applies here, right?

Because we don't know with certainty how hard it is to cheat at the lottery. Of course if it comes out that some people are cheating we'd know that that's at least one way to cheat.

So you have a large unknown variable there in your assumption.

How does the religion example apply? I don't understand that so much.

The religion example applies because it is the same phenomenon, but the one that has the most implications for how to live.

The phenomenon is that there is really strong evidence FOR something and really strong evidence AGAINST something. They are incompatible. Yet according to Bayesian reasoning you are like an ass stuck between two bales. You can't make a rational choice as to which is RIGHT, so people wind up picking one based on other reasons.

If you also consider Pascal's wager then that increases the weight on one side. In short you are between two massively compelling pieces of incompatible evidence.

Assuming that the world exists roughly as we understand it and is coherent and rational, evidence reflects the state of the world. There cannot be strong evidence both for and against something, because all evidence is derived from the same, single, truthful state of the universe. You're misinterpreting one (or both!) pieces of evidence - maybe they're actually weak pieces of evidence, or you're mistaken about one (or both) of them.

The idea that the world is inexplicable by evidence and you must make irrational choices to satisfy your own desire for an explanation is contrary to both science and religion.

This is all before you get to Bayes - this is just about understanding the nature of the world and the nature of evidence and experience.

> There cannot be strong evidence both for and against something,

Of course there can be. Multiple cases exist. Light as a particle and light as a wave.

That's a great example of misinterpreting evidence—to see the evidence that light is a particle as strong evidence that it's not a wave, or vice versa, would be a misinterpretation of the evidence based on the incorrect assumption that it can't be both.

There is a single, coherent thing that light is, which produces the evidence we see. Some of that evidence is consistent with being just a wave, some is consistent with being just a particle, and yes, it's also strong evidence that it's not just a particle and not just a wave. We don't know all the details yet; "both a wave and a particle" is a pretty good approximation given our current level of knowledge, as is "sometimes a wave and sometimes a particle."

But picking the description "It's either just a wave or just a particle depending on how you feel about it, and there's no evidence that will help us because the evidence is contradictory" is wildly wrong and unscientific.

> Pascal's wager

Have you considered the inverse Pascal wager?

That is: the "true" God is a trickster (maybe Devil if you prefer) who has planted Pascal's Wager into our culture. The Trickster God sends people who believe in Pascal's Wager to hell, and sends those who understand this inverse to heaven (including Atheists who were unconvinced of Pascal's Wager). After all, what use does a Trickster God have of people who are so easily tricked?

Now, I'm personally religious. But this isn't here nor there. The point is that Pascal's Wager is an interesting thought experiment, but its often taught from only one point of view. Its not really built upon a strong philosophical foundation at all.

Ultimately, basing your faith on what amounts to a philosophical question from the 1600 without considering its counter-arguments (grown over the many centuries) is a bit of a folly. Of course there are counter arguments, Pascal's Wager is incredibly old.

There is no strong evidence for God except a bunch of belief from people.

That is not quite true. You might think it's not "indisputable" however there is evidence that corroborates many of the Biblical accounts as studied by academics and scholars [0]. Much of this is archeological, statistical, along with a significant amount sociological support as well. There is generally still some level of "faith" involved in religious beliefs (obviously), however it does not mean that those that hold them hold no scientific evidence at all.

It's possible that you have researched for yourself and have some evidence for your "beliefs" presented by your statement. Many people (admittedly on both sides of the fence) have not done much research in this regard, but in my opinion is a valuable exercise, so if you haven't I'd highly suggest it.

[0] The Case for Christ (Lee Strobel), along with The Case for Faith and the Case for a Creator. He was a former atheist who initially set out to prove God did not exist after his wife converted. This is certainly not an extensive list of resources out there, these are just some that I remember off the top of my head.

Oh I do not doubt that some of the stuff in the Bible is true. It is a book about genealogy of course. But that does not prove God in any form. It just proves that some stuff in a book that people wrote happened.

> He was a former atheist who initially set out to prove God did not exist after his wife converted.

Well that's the thing, you can't prove something exists. It doesn't work that way. You can prove it does exist though, and that's how I'm interpreting your statement.

Since you mention Pascal's Wager one could also mention Douglas Adams "'I refuse to prove that I exist,' says God, 'for proof denies faith, and without faith I am nothing.'". And so if we did prove God exists then you wouldn't need faith which is what is being asked for.

Evidence that corroborates Biblical accounts suggests that those accounts have some basis in fact. Such evidence is orthogonal to the question of whether or not any deities exist.

Strong evidence? Beyond it simply being a collection of historical accounts by historical people, of varying accuracy?

Reply on news.ycombinator.com