The fact that productivity increases in big cities is not new finding. Economies of agglomeration is well studied subject. https://en.wikipedia.org/wiki/Economies_of_agglomeration
This study is modelling just one of the reasons why it is so little better.
The paper's primary policy points appear to be, using quotes from the paper:
1) CNR workers work better in "cognitive hubs" that specialise in this type of work.
2) "The main culprit of spatial misallocation is the existence of large occupation-specific externalities combined with potential distortions due to land use regulations."
3) Small towns are misguided in trying to lure "tech" to their cities, and should instead find a non-CNR industry to concentrate on that is not already concentrated in a specific city. ("the economics of the problem suggest that, with the appropriate transfers, small industrial cities in the U.S. should attract non-CNR workers and not try to become the next San Jose. The concentration of CNR workers in a few “cognitive hubs” should be encouraged, not scorned.")
4) CNR cities should be subsidizing non-CNR workers outside of the hub, at a UBI (wealth transfer) to non-CNR workers of about $17,000 a year per worker. ("CNR workers, who earn substantially more, end up paying a base transfer of $16,856. One interpretation of this base transfer is that of a 'universal basic income' paid to all non-CNR workers.")
We can distill this further, to:
a. In general, CNR workers work better around other CNR workers due to positive externalities of location. I think this is tough to deny, with SF/the Valley for tech, NYC for finance, LA for entertainment, Shenzhen for hardware manufacturing, et cetera.
b. We should be removing any restrictions that stop this cycle from growing, from land use regulations, to stopping incentives and funding to small towns to create their own 'tech hubs', to just straight paying people to leave the city that are not part of that industry and stay out. It's the latter part of this everyone seems to be latching onto, however it's pretty common in economics to just model incentives as direct quantifiable costs.
This also explains what is happening in the Bay Area right now:
"Otherwise, cognitive hubs might use other indirect means of pushing out non-CNR workers such as, for example, housing supply constraints, zoning restrictions, or a lack of investment in transportation networks to aid commuting. Such efforts can generate occupational polarization across space without Pareto gains for all workers."
Think of it this way (very much simplified):
If cities A and B create are producing 10 units of value each and small towns C, D, E and F are producing 3 each and you relocate all the 'important' (again, simplifying) people to A and B so they produce 20 units of value each while the small towns drop to 1 each you've according to the paper optimized and reduced the waste. After all 43 > 29. But from society's view this is a net loss. It would be much better to see how the small towns go from 3 to 5 each in order to reduce the gap.
Of course this is super simplified and in the end wealth distribution is a complex subject but this paper strikes out for me.
Inequality is best addressed by free healthcare, education and direct money transfers, not by slowing down things that create wealth and increase efficiencies.
That said PG did more to lift up people to riches than the bulk of the paper penning econometrists.
So, start-ups are fine, but making it harder for people from smaller towns or rural areas to get proper medical care, legal representation, education and so on because of perceived economic benefits of letting them rot in favor of improving the lives of those in the cities is missing the wood for the trees and will do a large amount of harm.
Providing free (but crappy) healthcare and education will not close the divide thus created.
So these things are not going to be solved without capable people in places that are behind already. You need good healthcare and good education at the same cost as in the big cities to make it make sense. If the price is free then so much the better but first and foremost this is a discussion about quality.
For example: in a certain third world country healthcare may be free. But all the good doctors have left for the United States because that's where they get paid the most. So now the locals have 'free healthcare' but it doesn't really help them. They probably would like access to better and affordable healthcare.
Didn't Paul Graham himself start a startup without VC money?
And there are ways of reducing the risk by not growing too quickly, building a company with actual revenues instead of burning VC cash.
The only thing reduced inequality would kill is the Silicon Valley variety of startup, which would be a good thing for everyone outside Silicon Valley. It would give sustainable businesses a chance to grow without having to compete with companies who burn billions of dollars to buy the market.