GraphQL man… the whole thing does this by design. Just in the last week I had to implement a custom retry function because the python requests one very reasonably doesn’t consider a 200 status code to be an error.

Yeah, this is the problem with frankensteining two systems together. Giving an LLM a prompt, and giving it a module that can interpret images for it, leads to this.

The image parser goes “a crossword, with the following hints”, when what the AI needs to do the job is an actual understanding of the grid. If one singular system understood both images and text, it could hypothetically understand the task well enough to fetch the information it needed from the image. But LLMs aren’t really an approach to any true “intelligence”, so they’ll forever be unable to do that as one piece.

Yeah, I don’t think the idea is a total non-starter, but I’d definitely like some details. How will this be limited to ensure it’s not being used by investors and house flippers? How will this be ramped down once the housing market settles to avoid it being permanently “priced in”? How will this be paid for and how much will it cost?

Unfortunately American political debates right now are more of a pissing contest about rally turnout than they are about actual policy details, because that’s what sways the voters on the fence for some reason.

Honestly I really don’t think that’s effective either. Giving people more money to buy something generally just means the market will respond by charging more money for that thing. The assistance will effectively get “priced in” given time.

It’s honestly the weakest part of the Harris/Walz platform for me. Trump plan is utterly insane top-to-bottom though, and they’re just using immigration as a scapegoat here, which is… something.