[<< | Prev | Index | Next | >>]

Wednesday, November 13, 2024

Some Ideas



Rehashing in short three ideas to improve the country (and world).


1) Health - How to quickly learn what works (for everything), and what doesn't, in practice

Setting: Often doctors and patients are on the fence on whether to try something or not, or to try this thing or that.

Proposal: Create a federal site/service where they can simply enter the options, and the site will flip a coin for them, and then track the outcome and publish the aggregated results (of all such micro-experiments) in real time.

Result: This provides a natural, huge, eternal study that would dynamically track the most pressing open questions (i.e., exactly the things doctors and patients would be on the fence about at any given time).

Misc: Because it always starts where a given doctor and patient are on the fence already, there is no expected risk or downside, and it is fully aligned with doctor/patient choice. As the site itself informs the discourse, the focus will naturally shift to different A/B choices which have already incorporated what has been learned. Random variation (in where a given doctor/patient pair find equilibrium) always explores the surrounding space, avoiding spurious drift. For questions of particular interest, the site could also facilitate double-blinding (e.g., by shipping either a drug or a placebo to the doctor/patient). This is a federal policy issue because regulatory hurdles and pharma counter-incentives would tank any private attempt.


2) Monetary Policy - How to make a truly stable currency

Proposal: Federal reserve notes (US Dollars) should be treated as if they were ETF shares in a national index fund (long only). This means the Federal Reserve should simply directly buy and sell publicly traded stocks, and potentially other hard assets (gold, land, etc), maintaining a roughly uniform distribution (holding at any given time roughly the same % of every publicly traded company and asset class), until one USD represents exactly one USD worth of assets (according to the open asset markets).

Result: This would make the money supply instantly and directly responsive to market demands (e.g. increased demand for liquidity leads to net selling leads to price drops leads to Fed buying those assets with new dollars to re-balance; and this reverses when demand subsides), would render major bubbles and bursts essentially impossible (because individual assets are priced in baskets of assets, so only relative pricing is even possible), does not grant special benefit to anyone when new dollars are created or destroyed (because they are not created from nothing--they are simply re-packaging bundles of existing, real assets), and would remove all human and political decision making (because the buy/sell formula is trivial and easily automated).

Misc: This is much like the gold standard used to be, minus the problems (gold's tenuous value comes from a combination of precedent and scarcity, and its scarcity doesn't scale with GDP). This is also much like the Fed supposedly works today, minus the problems (bonds are not a suitable asset--the circularity untethers the dollar from reality). It could theoretically be achieved gradually over time simply by shifting the fed's balance sheet toward hard assets.


3) Election Integrity

Proposal: Provide everyone a digitally signed receipt of their vote which they can check against a publicly posted list. Use randomly assigned unique IDs for each vote, and fill in alternate options with already-cast votes (previously used random IDs, as cast) to obscure which particular vote was theirs.

Result: Everyone can confirm online and after the fact that their own vote was counted properly, and can prove if it wasn't (with a digitally signed receipt that will conflict with the public list), without revealing to anyone what their vote was (because the same receipt happens to prove some other people's votes too--and you can't tell whose was whose). Anyone can confirm that the votes were correctly tallied.

Misc: The user, or software (such as on an open-source phone app), would visually (and/or digitally) verify that their pre-generated unique ID was copied to the right slot at voting time. Beyond that they have an anonymised printed record. A list of who voted should also be either public, or easily independently auditable under NDA, with the size of the list published in the latter case, to match (just the size) against the list of votes (each vote being an Anonymous-ID/Voted-Option pair). The pool of already-cast votes would be primed initially with a single unique ID for each option on a ballot. This can work in an entirely in-person voting system with voter-ID if wanted.




[<< | Prev | Index | Next | >>]


Simon Funk / simonfunk@gmail.com