jasonmarks_

- "I've already been hitting timeouts on NWS gridpoint forecasts.”

Whattttt? This is bad behavior on your part as a redistributor. It is more polite to do a bulk download (NDFD) and iterate against that directly.

Zero money take: quantum computing looks like a bunch of refrigerator companies.

The fact that error correction seems to be struggling implies unaccounted for noise that is not heat. Who knows maybe gravitational waves heck your setup no matter what you do!

That is another viable path, no more no less.

These look like staging MVP releases with a full rollout planned for the future. They are only including a few parameters at every 6 hours which is barely interesting to anyone with their feet on the ground.

> Don't get me started on the typos in Lost Art Press's _Virtuoso: The Tool Cabinet and Workbench of Henry O. Studley_ --- they mis-spelled the subject's name on the inside cover and duplicated one photo, so a pair of flat pliers is shown twice and there is not detail photo of the iconic twin pair of jeweler's pliers, and didn't do a "cancel" reprinting that page as any reputable publisher would.

I am not familiar with those books or their content but that definitely reads as if the intent has been substantially changed. A typo 100 years ago might have been a letter off in the type setter; the typos these days are rewrites!

> I've _never_ read an ebook w/o finding at least one typo

This is unacceptable. Typo's are not just aggravating but as they accumulate they begin to veer towards mutating the authors original intent.

I own last years Kindle Scribe model and enjoy reading with it. Technically, I probably just like e-ink devices and this was my first e-ink purchase. The Notebook's (now Workspace?) are a compelling experience but it is unclear how the syncing feature protects data privacy. Pen and paper still has a cozier vibe when trying to keep drafts of ideas secure.

Two critiques: - Kindle would be a much better product if kindle.amazon.com took me to a dedicated UX that is not washed out by the e-commerce bloat that currently surrounds it. - You have to carefully purchase Kindle editions of books. There are definitely Kindle edition books for sale that are digitally scanned, imported, and compiled as a Kindle edition with no proof reading having occurred leaving you stuck with typo riddled messes.

More objectively it reads as if none of the models performed well outside of 24 hours with a significant uptick in inaccuracy after 72 hours.

> I find it interesting that they quantify the improvement on speed and number of forecast-ed scenarios but lack details on how it results in improved accuracy of the forecast per:

Definitely. Training on the historical data creates compelling forecasts but it comes off as a magic box. Where are the missing physics for the high performance cluster?

> Im pretty deep into this topic and what might be interesting to an outsider is that the leading models like neuralgcm/weathernext 1 before as well as this model now are all trained with a "crps" objective which I haven't seen at all outside of ml weather prediction.

You are a bit misleading here. The model is trained on historical data but each run off of new instrument readings will be generated a few times in an ensemble.

> Global Forecasting System continues to get worse

What do you mean?

> Punishing neighbor properties for not getting that one deal is an interesting but not ideal solution.

It could very well be that it is less than optimal to compare lots intended for residential with lots you would develop a convenience store on.

> So this is all about getting the town to assess lots at higher valuations and this way raise tax revenue which in turn will raise town’s ability to spend for itself.

That is one angle of view. Alternatively, you could be encouraging vacant lot (or equivalent structure) owners to sell if neighbors are improving properties while they are not.