I've had a project bouncing around in my head for about a year. It's atleast a little naive (but hopefully not fatally so), and has some significant hurdles for it to actually work, but if it's actually possible, it's worth doing. I invite people to comment below about why it can't or won't work, or how to ensure it does. And I'll need help from people more qualified and experienced with the different areas than myself.

The scenario

This project isn't intentionally political. It's about combatting incorrect information, half-truths, and outright lies. These things happen everywhere, but they seem to be super-concentrated in the political sphere, so my examples are political.

When you watch a (for example) Presidential debate, both candidates will quote various numbers as facts. Some of the numbers might seem contradictory, sometimes so much so that they can't possibly both be true. Sometimes they're vague enough that, if taken in a strictly literal sense, they can both be true, and the speaker is only guilty of implying a different meaning for the given fact, but not technically lying. After any such debate, the 'fact checkers' come out in force, pouring over every statement made by either speaker, tearing down any incorrect fact, and pointing out any intentional or unintentional inference that would have implied a wrong fact. This scenario leads to two problems:

• The fact checkers, when reporting on the accuracy or inaccuracy of what was said, are just as capable of bending the facts as the original speaker
• Most of us aren't important enough to have fact checkers

The first issue means that, if you didn't trust what one debater said to be true, the same thought process will likely make you doubt what the fact checker says. The second issue is more important to me, however.

When having an honest debate with a friend, colleague, acquaintance, or family member, the interesting part of the discussion is over the differences in opinion. This is to be expected, and everyone is entitled to their own stance on what a given fact means, and what should be done about it. Person A may say that a president's foreign policy is problematic because there were X attacks on US installations during his presidency, while person B could say that X is a lower than normal number given other events in the world sphere. This, in my opinion, is a constructive debate.

In the above scenario, things are drastically different if you can't agree on what the facts are. If you and I are discussing how a new tax system is affecting the economy, we can debate over whether or not the unemployment rate is unacceptably high. The whole debate falls apart if we can't even agree on what the unemployent rate is.

It all boils down to the term fact. That small atomic bit of information, that can be true or false. Do they really exist? Are there facts that are either 100% true, or 100% false? Or are all facts "sorta" true, and depend on how it's worded?

I think that, if stated properly, and with enough specificity, there can be facts, and they can be true. I also think that most interesting pieces of information rest on less solid information, but are still worth stating, if the contextual information is provided. This is where Alethia comes in.

Simple description

Ideally, the Alethia project would be a website where you could look up "Are humans causing global warming", and you'd get a page called "Humans are causing climate change". That page would be a factoid (oh god, I need a better name). The page would have a reliability score on it that gave a very rough estimate of how solid that conclusion was, based off the other factoids it depended on. These "foundational" factoids would each have a reliability score, which is how the main page would calculate its number. This "chain of trust" would continue all the way down to irrefutable facts (if those exist). Facts which are not controvertial at all, and have no reasonable complaints or issues against them.

If such a dependency tree could be created, we could break down a big issue into much smaller pieces. Each factoid should link to all its baseline factoids that it depends on. Someone with knowledge of the area in discussion should be able to look at one factoid, and, assuming it's dependencies are reliable, he/she should be able to make a decision on if the conclusion reached by this factiod is correct, overreaching, incorrect, misleading, etc.

If an extraordinarily complicated question like "Are humans causing climate change" could be broken down enough, then someone like me might be able to comprehend the links in the chain. If I looked over a couple links, and someone else looked over a couple more links, then with the multiplicity of the internet, these conclusions could be verified. Open-source software benefits from the same effect, as the quote "given enough eyeballs, all bugs are shallow" illustrates. If the system were granular enough, and there were enough people looking at it and editing it, it might start to gain broad trust. If that happens, then we've elevated the debate to where we're disagreeing over what the facts mean, not what the facts are.

Isn't this peer review?

Yes. This is basically internet-scale, granular peer review.

You're just lazy!

Yes, but this isn't just because I'm lazy. I could do my own research for whatever topic I'm curious about, but it'll take a while. I could even blog about each thing I research, so others could benefit. And people could even contradict me if I came to the wrong conclusions, and I could correct the posts. This isn't good enough because: First, I'm lazy. Second, my blog won't become something that people on all sides can trust. It won't be a place that, when you read something that goes against what you wanted to believe, you can't just ignore it because of my bias.

We need a site that has a robust conflict and verification process so that, if you see something you know is wrong, you can correct it. And if you see something don't like, but only hope it's wrong, you're forced to consider that it might be true, because all past objections (which must be visible) have been refuted with evidence.

Hasn't this been tried before?

Maybe. If it has, and it failed, maybe there's a way to try again and succeed. If there's good evidence that the problems are just way too compilicated and can't be broken down into "factoids", then maybe it's not worth doing. If it's been tried and is currently working and growing slowly, tell me where, and I'll help out that project.

You're crazy

Also maybe. If there's any chance of this working though, I figure it's worth a shot. Please comment below about:

• Why it won't work
• Why it didn't work in the past
• Why I'm oversimplifying
• Why it's a hopelessly naive solution
• Why it's hubris to assume that a problem as large as global warming, which many people devote their entire careers to solving, can be broken down into a small enough "factoids" that one internet denizen can understand it all
• What would you change to give it a shot at success