A Big Board of Problems

Would be valuable to both supervisors and independent researchers

One of my running theses on this blog is that there are highly skilled people that are capable of independent science. But, the biggest problem is that they don't know where to find problems, don't know where the frontier is, or how to find supervisors to guide them. I myself found problems by cold emailing and networking with hundreds of people, but this is inefficient.

One the other hand, academic institutions have no money but lots of ideas they want to try.

So, we need…

A maintained, interactive, big list of open research problems amenable to independent researchers. To get a better idea, I’ve wrote a spec of what this might look like

• I have a prototype here, which has been used by our discord. I am really keen to get open problems from other universities, FROs, government science agencies like ARIA, science aligned non profits, and other research organisations on here.

• The target audience should be people with degrees, some research experience, but didn't want to go into full time PhDs / post-docs.

• Each item list should be 'sponsored' by an active supervisor in a sufficiently 'prestigious' institution, to give the list legitimacy, get people talking about it, and critical mass.

• If I had more resources, one job would involve continuously speaking to universities, FROs, pro-science philanthropists, and other research organisations to contribute problems to a big board of projects.

• Independent research problems also need to be filtered such that they are amenable to 'decomposition' where independent researchers can contribute separate chunks. This can be hard because problems often have a large degree of entanglement.

• The problems should also aggressively be selected based on how active the supervisors are, having famous academics who are too busy to help new researchers is a failure mode.

• There needs to be technology to ensure a strong user experience, such that we can match people's interest and skill to appropriate projects. I think LLMs would be great for this.

• The items in this big list ideally should be more than a list. It should be interactive, such that supervisors can set 'tests' for prospective collaborators, like a math challenge. It should also contain ways for people to 'bid for supervisors' by submitting tests of skills.

  • There should be interactive tests so that supervisors can chose promising candidates to work with.

• Some projects that I think are amenable to independent science are the following. But we should have a systematic way to identify many others.

  • Physics / Math formalisation projects like Lean and PhysLean

  • Mass collaborative math problems like busy beaver

  • Better infrastructure for scientific computing

  • Mass paper replication / community peer review - why not outsource paper replication to bored independent scientists.

  • DIY methods to set up home labs - can we make a $80 scanning electron microscope?

  • 'Field translation documents' - we all know humans have made progress mixing different fields together, but its often hard to learn the language of one field efficiently.

  • Data set collection for tricky things.

Acknowledgements

Thanks to Jess Carr for the discussions.