>I took on silly amounts of work for absurdly low amounts of money.

I think you might have stuck longer to the lower rate + spec work. The idea is to start low and ramp up very very quickly. Infact initially you should try to double your rates, and adjust(A/B test!) your rate according to response rate.

Also i think rate is not a static thing. It depends on how much work is on your plate, or how less. It also depends on the market. So when tomorrow a lot of these startups go belly up and billion dollar valuations runs dry, so might your work. You will have to adapt accordingly.

If you don't have enough work at a particular rate, there's no harm in claiming a lower rate as a few test data points. You can always bill some hours you might not bill on the higher rate. For eg. sometimes i leave off some of the research hours when i'm billing a higher rate, but i bill them when i bid a lower one for whatever reason.