Human powered blockchains?
One of the problems that conventional blockchains suffer from is that they’re expensive to run. You’re essentially burning electricity to protect a blockchain and as the difficulty rises you’re forced to expend even more energy to achieve the same result. Then there are looming increases in algorithmic efficiency (see ASIC boost) and hardware speed (see ASICs) that make the process harder to maintain in a fair and decentralized manner.
I was thinking about CAPTCHAs today and maybe they would work as an alternative to hash-based proof-of-work. You see, human cognition is much, much harder to improve than processors are, so essentially any proof-of-work based algorithm that relies on human abilities to run can only be improved so much. I believe this is quite a significant observation when applied to blockchains because there may be algorithms out there that are capable of transforming a block hash into an arbitrarily complex abstract reasoning problem with a publicly verifiable solution.
As far as I know: this category of algorithm has had little research done behind it as traditional CAPTCHA algorithms are designed to be solved in under a minute so the problems that they produce are simplistic (often simplistic enough for computers too, sadly.) But if you remove that limitation and allow solutions to be solvable in arbitrary amounts of time then you might be able to produce CAPTCHAs that are only solvable by humans, yet whose result can be verified based on a strict problem definition.
If such an algorithm were to exist then we could produce blockchains that were inherently more decentralized than anything that currently exists, that would solve the centralization of mining problem, that didn’t suffer from huge amounts of electrical consumption (i.e. cost), that was difficult to make parallel (so parallel resources wouldn’t matter as much), and was arguably more secure than what we currently have (baring the creation of advanced AI with human-level intelligence and reasoning ability.)
Additionally, more conventional CAPTCHA algorithms can already be used as a practical time-lock encryption algorithm without the current issues of having the encryption broken much earlier than expected (which will always be an issue with current advances.) Meaning that the DAC that I described in the timechain white paper is not only practical today but also potentially much more secure (since the human brain is harder to optimize.)
Will human intelligence be the next arms race?
A brief sketch for verifiable complex CAPTCHAs:
- You would define a portion of bits in the block hash to represent the initial setup for a complex, 3D world.
- The remaining portion of the bits would be reserved to describe a desired outcome after manipulating that 3D world.
- The corresponding solution would define a series of arbitrarily complex manipulations to those 3D objects to produce a desired state.
- To verify that a solution is correct requires replaying those manipulations against the starting state to see if it produces the end state.
- The game would be defined based on manipulating aspects like color, size, definition, ratio, placement, etc to stress different aspects of human cognition like pattern recognition, numerical reasoning, linguistic reasoning, and abstract reasoning abilities.
That’s what I have so far. It’s not much but its a start.