Eighteen open problems

Ideas I have not built, written down so people can tell me why they are wrong. Idea one is already running at taxslop.com, so it is not here.

1. A professional network that bans automated outreach

   Can a members' network make "a human wrote this" the whole product?

2. Getting two people's AI agents to hand over context

   What replaces emailing a summary from your agent into someone else's?

3. Capturing what an owner knows before the business closes

   Is tacit operating knowledge transferable out of the person who holds it?

4. Making a digital footprint unpredictable rather than private

   Is poisoning a profile better than trying to delete it?

5. Files that stop working when their surroundings change

   Which environmental signals are trustworthy enough to justify destroying something?

6. Building restraint into agents at the level of means, not goals

   Is there an artificial analogue for the thing that makes a person stop?

7. Communication infrastructure for a world of personal agents

   If agents talk to each other, why are they doing it in English?

8. Which dead industries can actually be reopened

   Did this industry die because demand vanished or because the costs stopped working?

9. The fork between detecting slop and using it

   Which side of your own product is the customer actually on?

10. Nobody is looking after the people being upskilled out of a job

    What does support look like when the job is gone rather than lost?

11. What people building new jurisdictions have to buy

    Which inherited function is both mission-critical and structurally underserved?

12. Things that are ordinary in one country and banned in the next

    Where is the gap wide enough to be a business and narrow enough to be defensible?

13. A physical limit on your own screen use

    What restraint can a person not undo in three seconds?

14. Sorting household recycling without reading the label

    Does anyone change what they throw away because an app told them to?

15. Removing a smell instead of covering it

    Can an unwanted signal be subtracted locally rather than masked?

16. The gap between thinking something and doing it

    Which single action can be inferred dramatically better than it is today?

17. Matching people on biology rather than preference

    How strong would the evidence have to be before this is not pseudoscience?

18. When the goal is fine and the means are not

    How do you specify a task so the unacceptable route is not available?