Ok, I'm about to sound hypercritical and rambly. What I'm actually doing is running through the mental models in my head for "how businesses work" and trying to come up with possible failure points and solutions. I think you've jumped ahead to the "how can we create this thing" stage without first running through the "how can we make this a viable, sustainable sytem" part.

Incentives: One thing you might run into is the typical problems with bartering systems - you have to find two people who have exactly equal knowledge that's worth the same to each other, that takes the same amount of time to teach, and is exactly the knowledge that the other wants - otherwise, the trade will not be worth it for one of the people.

One way you could fix this is to have a "time currency" - If you tutor someone for one hour, you than have one hour that you can use to "pay" another tutor.

The problem then becomes quality of the tutors - because there's no form of quality checking in the system/pay for performance, so there's no incentive to improve your tutoring. One way to get around that would be to make a rating system that tied in with the "hours pay", so that for instance if they get five stars, you multiplied the hours by 1.25, if you got four stars, you multiplied by 1, three stars, by .75, two stars, by .5, and one star, by .25. This would incentivize people to actually be good tutors so they got "paid" more.

Network Effect: This is a network, so the typical chicken and egg problem occurs with the network effect - it only works if you have a ton of people who can tutor in a variety of fields, but you're not going to attract that ton of people if there's not already tutors there that want to tutor them. There's a few ways to deal with this, and I think that you should use all the strategies if you want this to work:

1. You can preseed the network. Get a bunch of your friends who believe in the project to agree to put in x hours of time per week, so that you start with a good amount of tutors and then draw other people in to what already seems to be a vibrant community. (this is how reddit did it).
2. You can create a prelaunch campaign and email list so that on launch, you have a bunch of people all coming at the same time, instead o a steady trickle of people who come, see that no one else is active, and leave. (most network effect startups try to do this.
3. You can add some extra value to the system that works solo - eg a self study curriculum that's on top of the extra tutoring system. This means that the initial people stay for that self study curriculum, then as the network grows, they start to see the extra value from the tutoring system (this is how tumblr did it).
4. You can start with a tiny niche, such as only people who want to learn and teach math related to rationality. This means you have to cover a much smaller base of what people want to learn and teach, and then you can expand slowly based on "peripheral interests" that you see a lot of your network has (this is how Facebook did it.)

Double Sided Market:

This is a two sided market, so you run into a similar problem of the network effect. You need tutors to draw in tutees - but you need tutees to draw in tutors. Now, this is a weird double sided market in that you're expecting your buyers are also your sellers... this might seem to solve the whole problem, but in my mind it actually makes it harder. You're assuming that there will be an exact symmetry in how much people are willing to tutor others vs. be tutored, but I would assume there will be a vast difference in how much people are willing to be tutored vs. tutor others. This in my mind is probably the biggest "what-if" in this scenario : If I'm right, and people are not willing(or able) to tutor somone for an hour to get an hour of tutoring, the idea becomes unworkable.

MVP: To me, going through the whole model, that assumption of symmetry between tutored and tutoring is the riskiest assumption, and the riskiest assumption should always be what you test in your MVP (minimum viable product). The simplest way to test this would just be to go on craigslist and put up an ad that says "Looking to be tutored in XYZ, willing to trade for tutoring in A,B, or C subject."

If you get say, 5 legitimately interested responses, that's enough to contact those people, talk to them about your idea, and see if they'll tutor someone. If they will, you can go on to the next step and start creating your launch list.

People: The first failure point that comes up in any startup is not usually getting a good idea, it's executing on it. The tone of your post seemed like it was saying "here's a cool idea I have" vs. "Here's something I'm going to do." But my model of this is that if you don't make it happen, it won't get done. No one else is going to execute this for you.

It would be great if LW could scrounge up enough resources to help people studying for the careers recommended by the 80000 hours. Starting with a list of the needed, but not covered in college, subjects.

List of participants:

What would be the advantage of bartering tutoring compared to the flexibility of simply selling and buying?

This was a startup; I think there was also one that focussed on language-swaps (google that for your local group). If it's not famous by now there are probably reasons why it didn't work. MattG below probably covered them.

Great idea! But I am crap at tutoring, any knowledge exchange would be very unequal.

I have seen that before, I think was a lw rationalist project but I'm not sure. From what I remember the web page was pretty dull, so maybe that make it fail(or prevent it to grow).

P.S. I will post a link If I can find the site.