Idea proposed here by Fernando Nieto.
Assuming there is a trustless BTC/USD price oracle, we can burn BTC to create a dollar equivalent amount of "space dollars".
E.g. If the BTC price is $20k, burning 1 BTC gets you 20k space dollars.
The resulting token is the equivalent of the USD: a stable unit of account, but a poor store of value.
If we assume that USD will lose value relative to BTC over time, the space dollar peg will be quite stable.
The key here is that because of inflation, more and more dollars are required to represent the same amount of non-inflationary value.
Consequently, this means that a decrease in demand (in non-inflationary terms) will correct itself over time, as the dollar value will eventually drop to match the demand.
The only way for the peg to lose value, is for demand to drop faster than the USD loses value for a prolonged period of time.
One way for an oracle to be trustless, is if everyone is the oracle: we make it a consensus rule.
Still, this isn't easy, because users have to get the BTC/USD price from third parties.
Consensus has to be reached on the time/price, with a window to allow for some flexibility (otherwise consensus may stall frequently).
The spacechain burning mechanism makes things more fair here, because only spacechain miners may burn, and they have to compete with each other for the privilege, so even if the oracle price is beneficial to them (compared to reality), the difference will become a fee for Bitcoin miners and thus be turned into PoW.
E.g. if demand for space dollars has grown by $19k and miners wish to create them, but the oracle is manipulated to think 1 BTC = $20k while it's actually $19k, spacechain miners will end up burning ~0.95 BTC and paying an additional 0.05 BTC in fees. In other words, even though the oracle wasn't accurate, the correct amount of space dollars were created for the correct price.