Suspicious Discontinuities (2020)

I cracked up when I got to the marathon example. When I ran a half marathon I realized about 80% of the way through that I was on track to finish under 2:30:00 and pushed myself to make it happen. I should have guessed that sort of behavior would show up in the statistics!

The UK tax system also has a bunch of unfortunate cliffs, and tapers that create >60% marginal tax rates and worse. There's a calculator here that illustrates it well https://tax-cliffs.britishprogress.org/calculator

The childcare cliff edge is probably the worst, but the personal allowance taper isn't ideal either as it's compressed over a relatively short income range

And of course all the thresholds remain frozen, creating plenty of fiscal drag on top.

Jim Roskind had a really nice talk [https://www.youtube.com/watch?v=_uaaCiyJCFA] at AWS Reinvent 2022 which highlighted the fence post problem. Basically, AWS was targeting P50 and P90 latencies in many services, and sure enough there were big clumps of latencies right under those target fence posts as engineers were basically gaming the metrics and nerfing the >P90 latencies to make it happen.

Beautiful! I noticed this on the chess ratings distribution on Lichess: players like to cross a nice multiple of 100 and make an extra effort to avoid dropping below if they can: https://imgur.com/a/Db7fQdX

Indian laws have the same too. For taxation 'surcharge' applied to high income groups, a patchwork called "marginal relief" fixes it [1], however, leaves ranges in income where 100% of incremental income goes away in taxes.

[1] https://cleartax.in/s/marginal-relief-surcharge

Another place where thresholding is misplaced is minors (below the age of 18 years) being subject to different laws (this is for India at least), e.g., seriously different punishments for crimes like homicide or rape. I have heard that crime groups then purposefully involve minors, push the blames on them if caught, and effectively get away for much less.

> A simple fix for the problems mentioned above would be to have slow phase-outs instead of sharp thresholds.

How about the even simpler fix of not having phase-outs? This generally works: the cost of a subsidy generally does not increase as the recipient pays more taxes, so the higher total tax paid by a higher-income person will easily pay for the cost of a subsidy. And giving higher-income/wealthier people the same subsidies as poor people may help them appreciate the ways in which the subsidies are helpful and the ways in which they suck, which can help the whole system improve.

I also appreciate discontinuities and while I won't comment on the data in the paper itself without cross-referencing, I will say that a couple of these examples hold true for me from applied observation over the years. When I was old enough to start caring about insurance for health and property, or became a parent and had to begin forecasting costs for college and what loans really represented, I began looking at observable data much differently. Working in the software industry, you begin to see the complicated systems at work at the C-Level, and the seemingly odd relationships with unrelated organizations start to become clear. Being an educated voter, a discerning consumer of products, and turning a critical eye on world news all require the ability to see processes, their patterns and the discontinuities within them. While there may not always be a useful explanation behind all of them, seeing them in the first place is essential to navigating so-called reality successfully.

Would test score problem be solved if teachers graded individual questions, not entire test?

yeah, slow phase-outs seems ideal- I have often wondered about that with gross and and taxable income.

Haven’t seen danluu stuff here for a while!

One way to reduce income may be to buy a bunch of stocks with high variance - for any that is in the red during the year, sell and rebuy them. Any that has a gain, leave them be.

The opening story is fabricated and/or bullshit.

Last year (2025) there was no limit on income for health insurance subsidies. That ended for this year, but last year there would have been no reason for anyone who knew what they were doing to try to lose money to drop their income (especially in the cited range of $48-55k/year).

That is the case this year, in most states (thankfully not where I live), but that's not what TFA is talking about.

Suspicious? It certainly makes me skeptical that the author has got the details of the other examples correct.

I never understood why taxes or similiar absolute points aren't gradients instead.

(2020)