But some/most guys want "one easy trick to get jacked" (cut labor costs without any decrease in output); many influencers (AI companies) are more than happy to hawk supplements that don't work and/or steroids that do work but also kill you (LLM-based automation)

I mean, the SPM exists, we already do measure poverty tagged to baseline expenditures, it's just that "the real poverty line is $40k, not $30k" and linking to a CNSTAT report doesn't get clicks.

I am cranky about this, tbh.

Its mean, fine, but that's actually a response that's in short supply tbh. We need to be meaner to cranks and return them to their rightful place of shame, not solicitous of them when they flatter our priors.

Its a real problem, but it is a real *substantially solved* problem: surf.econ.uic.edu/wp-content/u...

You don't get credit for recognizing what's been obvious for decades, not doing one (1) google search on previous work on poverty and then doing some dodgy math.

"but the crank who did the math wrong had a point and gets at something in the vibes" is rather how we get *gestures all around*

The objection is not to a poverty threshold based on expenditures on a minimum consumption bundle -- that's what the SPM does -- it's the ignorance of the fact that this is a known problem (hence why the SPM exists), and the shoddiness of trying to "solve it" using napkin math

He didn't provide any insight, he rediscovered a shitty wrong version of the SPM and did the math wrong. People should read the literature!

This is part of the "so close to an insight" -- the supplemental poverty measure is constructed kind of like this, and sets a threshold based on 83% of the median food clothing shelter and utilities expenditures in the CEQ microdata. Its about 40k in 2024.

Apple pie you probably do day of, but I'm not waking up at 5am; day before has minimal quality dropoff

Edge case is pies though, right? If you're doing pumpkin or sweet potato pie, you actually want to make that day ahead.

This is the issue with the genre of "the poverty line is too low" -- fine, maybe it's 1.5x the SPM threshold or whatever. When you adjust it back that just means that *even more* people were poor according to that anchored threshold in the past

(guy standing up meme) saying true facts about the world is an expression of values, as is rejecting true facts about the world, and the former set of values is better

You are probably not going to write down the correct model. You might not even know you are writing down a model (implicitly)!

Much easier (I mean this sincerely) to just pay the fixed costs of figuring out how to access the microdata

There's also a fun lesson here for the microdata heads: trying to answer these types of questions with just aggregate data is *way harder* than just using appropriate microdata. If you have the microdata, all you need to do is calculate some summary stats, if you don't you need to build a model.

The other fun thing is how much dude *almost* gets to some fundamental insights -- he comes close to learning about how to capture relative poverty, the differences in the distributions of consumption vs income, savings rates, etc. but keeps trucking ahead with bad math until he lands in absurdia

There are regional price differences applied under the hood on the SPM, but this is an active area of research, definitely can be improved (as the various CNSTAT committees have noted)

The SPM threshold for a family of 4 (as in OP) is around 40k -- this ends up being about 1/3 of median income for this family size, which is a pretty typical way of capturing relative poverty in the literature.

This is not what tech guy is saying,l (because that would require subject matter expertise), but there is a smart version of this where you assign lifecycle and family size specific poverty thresholds and do redistribution based on that -- which probably does back you into subsidizing child care