2 facts for the 2020s
Here are the two concepts which I think will both be key for navigating the 2020s. I might be wrong about both, but the exact way in which I’m wrong is very important to analyze.
- The most important difference between the existing central banking system and a system like Maker is that when traditional central banks make loans, the interest on the debt causes there to be more debt than credit. In a system like Maker, the interest on a CDP’s debt is actually created and released into the economy via the MKR buy/burn auction in real time, such that the total debt equals the total dai that actually exists (modulo constant-factors like the fact that the debt auction can only be ‘poked’ at finite time or quantity intervals, or other factors of this form).
- Corrolary: The issue of whether inflation is necessary for economic growth is confounded by the fact that inflation is necessary to allow the economy to pay down the debt at all. Of course there will be credit crunches that require ever-increasing loans if total debt is grows every larger than total credit! Inflation might still be necessary to cover e.g. lost or ‘hoarded’ credit-units, but that necessity is currently used a mask over a perpetual debt treadmill.
- Both Bitcoin and Ethereum can only compute finite functions. Neither is actually turing complete and neither is your computer. Some say it has to do with statefulness but that is only part of the equation. The most important aspect of money-programmability is the composability of behaviors that are written independently. In Bitcoin composability comes in the form of “all inputs must validate for the transaction as a whole to be valid”. In Ethereum, REVERT reverts all state changes, even those made by earlier sub-calls.
- For Bitcoin (with no block/tx size limits) to be as flexible as Ethereum, all needs is the ability to “check” other inputs and to “set” subsequent outputs in the same transaction by examining them and verify conditions such as “the first output’s script equals the first input’s script”. It would be a very different programming model, but most of the same dapps would be possible and they would be nearly-optimally-scalable by default because UTXO validates in polylogarithmic parallel time.