Cost plus have been disfavored in the FAR for a long time. That doesn't stop contract officers from writing them. The biggest thing, imo, is that congress needs to improve its oversight of ACAT I and ACAT II programs and stop waiting for them to be a hundred million overrun before yanking them.
You have to realize that it isn't just about saving deer. If you hit a deer at speed, it is going to do massive damage to your vehicle and potentially cause significant injury to the passengers, up to and including death. Overt fatalities are reportedly rare, like 0.1% or something, but the injury rate is closer to 5%.
Someone very likely 3d printed an AR lower (unserialized, therefore a "ghost gun"), unlikely it had an autosear. The news and police typically call disfavored semi-auto weapons "assault rifles" and things that are fully automatic "machine guns".
Most of what goes on on most bases is as unsecret as it gets. The rooms where sensitive things happen have security measures in place and they keep you from bringing your phone in there. Usually a little locker is provided where you store your phone while you are inside the protected area. In some very sensitive places there may be additional active and passive measures but usually people with access to those areas police themselves.
Like a lot of data collection, the meta data gets you pretty far. The who's and where's can be hoovered up with tracking data buys and essentially get you a nice list of who has access to certain areas, which can be used for targeting individuals.
That is largely an artifact of the source data, sadly. I place the dots where they were found, if the dataset has that information, otherwise I georef them back to the "findsite" in the data, which aggregates many finds at one place. A lot of these were found in the 1800s and just dont have good find lla data.
Ah, I was referring to the zoom-level-dependent clustering, which I find makes it hard to see the distribution of points when zoomed out all the way. There's still quite a bit of detail even with many of the points sharing the same location.
But yes, the runbook in the project gives the llm instructions on how to use the scripts and what to modify. I let Claude code read that and tell it to work on a province. It runs a small segment and analyzes the results until it hits 1-2% error with no systemic errors. If it can't get all the errors out then I have it switch to using gemini-flash-lite-latest instead of 2.5, which costs slightly more but performs much better. Basically Claude code runs a self governing loop with my oversight mutating the prompts and data inputs to extract all the names.
The actual code is here https://github.com/metiscus/roman-names/ and it is licensed under MIT, while the datasets are slightly more restrictive, and mentioned in the repo.
Yeah, I thought about that. The issue is that the date density is poor already and the ranges are pretty broad. Someday I'll give it a try, but the search interface can limit you to date ranges for now, so the infrastructure is all there.
reply