Thanks for the feedback. The description is cut short because of the 2000 character limit here on HN. I did post a follow up comment with the rest of the description, but it has now fallen to the bottom of the comments so nobody seems to be seeing it :(
The basic idea is to bring technology to the certification process in education (a huge industry). Right now all the new education institutions (and a lot of the old) have no way of really proving to the world that their graduates have learnt something and have valuable skills. Employers see a piece of paper from some institution and wonder what does the candidate really know. This is what the idea tries to solve. Now to your specific comments.
Testing and certification is the most economically valuable aspect of education (at the high end the most valuable test a student can take is getting accepted to an elite institution (I'm using "test" broadly here)).
Exactly. The value of an elite institution is not in the actual education you might get, but the hurdles needed to overcome to get in. It is an incredibly crude proxy for measuring what an employee will actually be like.
Learning itself can and has scaled quite easily thanks to the internet. But testing and certification hasn't really yet.
Yes. The reason why is that there has been no way of preventing gaming of the testing system. What you want to measure is if a person really know their stuff, or has just crammed it all in before the exam. My idea is break this link between cramming and test results by preventing cramming.
There is a huge opportunity to build a new authority for testing/certification/credentialing.
Yes there is and once built you will have a very strong network effect that will make the business very valuable.
...In other words, it's not the algorithms themselves, it's that you're disciplined and smart enough to figure out how to learn them.
This can actually be tested too. Once you make a test unable to be crammed then you can start to do the sort of tests that measure real on the job performance.
Another strategy would be to go after the higher end, and try to become a trusted authority for certifying employee skills. I think this is better because this seems like the kind of thing where it's easier to go from high-end to low-end than the other way around.
Yes this is my thinking too. Start in a narrow niche and become the authority for certifying the skill set of people in that niche. If the testing authority say you really know skill X then as an employer you will believe this is true. You can add all sorts of layers to this where you can rank a persons skill as beginner, intermediate, expert, etc. Once you have a strong foundation in a niche then just expand into related fields.
Certifying knowledge workers would probably look less like giving them pop quizzes and more like conversations with them, of else just tracking their work closely and assessing it.
Because the tests can’t be crammed the tests that would be given would by design have to be a much closer match to real work than a typical test. You can make the tests as rigorous and challenging as the market wants. If people want to be certified that are lateral thinkers that can solve real work problems then that is exactly what can be tested. You don’t want to test people on things that they can google in 30 seconds, but the sort of thinking and problem solving that make a real difference in the workplace.
To get it off the ground you need to decide who is your customer - the worker or the employer?
There are two customers - the worker and the education provider (see my follow on post for more details). It would be easier to go after the workers at first, but the educational providers would be where the real money would be.