Set the thresholds using a countrywide baseline (1.0 = countrywide median household income) and adjust according to the median household income of the city the job would be located in. Also adjust thresholds based on the profession.

Those same companies then claim position is small town x, but that office magically closes when H1-B arrives.

A better way might be weighing based on the applying company. Give a weight based on the size of the company and another for the type of company. These weights would be assigned based on the top-most owner(s) and not the applying companies themselves.

Also add penalties for larger companies acquiring smaller companies with H1-Bs that were hired within the last X number of years. Not sure what penalties exactly. At least financial to fund the system, but probably to the weighing as well.

That could potentially be solved by having different buckets for different regions and industries? Seems like that'd be a much fairer system than the current lottery one.

Different regions can be easily gamed by opening PO Box offices in all the regions.

Well, then you check tax return of employees -- if they don't actually reside in the same region, the last person who touched the immigration form is getting a date with a rattan stick.

Fair is an odd word here, the system should work for the greatest benefit for Americans.

I think a wage threshold and weighting would be best, but others think a controlled random mixing of labour markets is better for the overall economy.

This exists. Here are the prevailing wage cutoffs for each experience level (calculated by the BLS) they use for "software developers" in SF, for example: https://www.flcdatacenter.com/OesQuickResults.aspx?code=15-1...

Regions and industries that pay less don't need more labor, hence the unwillingness to pay for it.