LLMs have lowered the bar for the unskilled person to create shit software. I have used Opus 4.6 on a number of projects, and it still spits out buggy, and sometimes, flat out broken code. I was actually surprised when it completely hallucinated the names of query params for an HTTP request in my code, when in the prompt I had explicitly given it the exact names it needed to use. I thought these frontier models were supposed to be game changing.

The query parameter issue is a pattern I see a lot. The model has thousands of examples of "how HTTP requests usually look" from its training process. When your input data conflicts with the pattern, the training data takes precedence.

Interestingly, the model doesn't "know" that it's ignoring you. From its perspective, it has retrieved a "meaningful" pattern—virtual parameter names that probably fit common conventions it saw during training. Your actual request simply... wasn't documented.

> LLMs have lowered the bar for the unskilled person to create shit software.

So? Demand the source code. Run your own AI to review the quality of the code base. The contracting company doesn't want to do it? Fine, find one that will.

Add another layer of jank to review the original jank? That doesn't sound like a very helpful solution. But the companies selling AI will love it!

Technical Supervision of the Investor is a thing, for a reason. The fact that IT industry doesn't have it is ridiculous.

And more importantly, think of the funding we’ll get

I hope that commenter was being hyperbolic. I have heard of headphones making visible impressions or “dents” in the soft tissue and msucle after long periods of use (Google Tyler1 headphone dent if you don’t believe me), but such a dent would disappear within minutes or hours. An actual deformation of the skull due to headphone-wearing would definitely be strange.

RuneScape private servers used to bring in tons of money. I helped manage one when I was in my early teens, and I can confirm the owner (who was only a few years older than I) was bringing in mid six-figures annually.

Then of course there was the rampant gambling. The founders of online casino Stake and streaming platform Kick both started their “careers” in RuneScape gambling. IIRC they invented “staking” which was a method of gambling gold against other players, before they were banned. But the gambling economy in RuneScape used to IRL mint millionaires for sure.

Staking was a built in feature of RuneScape in the duel arena before it was removed. They did not "invent" it themselves.

They did run a dicing clan as well, FWIW, although I doubt they were the first to do it.

Example?

I think it’s critically important to mention that the VA provided all this data to Ancestry.com years ago. According to the newletter op linked, Ancestry.com charges $300/year for access to this data. This unfairness is what prompted the lawsuit and ultimate release of data.

Indeed. This is ABSOLUTELY not the first time we’ve dealt with a government agency (at the local, state, or federal levels) providing a copy of a public dataset to Ancestry.com and not to the general public. Our taxpayer-funded data keeps ending up solely behind a $300/year paywall. It’s not fair.

(Also, the stripped-down version of BIRLS that has been on the Ancestry website for a while now is much smaller and older.)

How recent was the last death of a veteran, given to Ancestry.com, compared to what your efforts have now exposed?

Ancestry has a somewhat smaller copy of the BIRLS database online, covering just the years 1850-2010 [1], and it seems to have been published on their website in 2011: https://www.ancestry.com/search/collections/2441/

Our data set from the VA contains data through mid-2020, and was turned over to us in 2022 after undergoing extensive double-checking by the agency, including through non-public VA sources, to confirm the veterans really were all deceased. There's a paper showing the agency's methodology on our site, which we FOIAed from them.

There are a significant number of deceased veterans whose data is *not* included in the BIRLS database, because they (or their family/heirs) simply did not have any contact with the VA concerning benefits in or after the 1970s, which is when the database was first starting to be built. That is, their files almost always still exist on a warehouse shelf somewhere, but they weren't active any time in the past fifty years so they didn't get pulled and indexed into the database. You can still make a FOIA request to the agency asking for one of those files, but the VA will have a lower chance of successfully finding the file, and it usually will take longer for you to get a response.

[1] 1850 is very likely an approximation. While there are certainly deceased veterans listed in the BIRLS database who had birthdates or deathdates in the mid nineteenth century and/or service in the late nineteenth century, they are relatively few. Many of them are actually veterans with likely birthdates or deathdates in the twentieth century whose data seems to have been initially recorded by the VA with a two digit year of birth or death or enlistment/entry, and then assigned to the wrong two digit prefix, causing an incorrect four digit year of birth or death or year of entry/enlistment into service.

In other words, the VA's historic data is very messy and is a great example of an actual Y2K issue.

Tremendous work on this, thanks for the work to unlock these records for the public.