Technical consulting for coding, predictive analytics, and optimization. Focused on public sector applications, mostly with police departments.

Based in United States
Filter
Exclude
Time range
-
Minimum likes
Working on updating the CrimJobs scraper today. (Sonnet is too expensive). One interesting thing building these bots, is that companies preventing them are equivalent to preventing Google indexing. So journal articles (and for CrimJobs the ACJS job board), this is just static content. There is no reason why you can't serve this to millions of requests a day to bots. For crim departments paying for ACJS, you should consider just paying for an institutional membership directly to CrimConsortium. It is better advertisement (we get a million views per month now). These also do not deter a sufficiently motivated user (GrokBot for example can easily get around it.)
1
1
4
203
Hallucination of the day in Probation Journal I have made the tools (when sharing these) as transparent as possible. So the NIJ article my tool originally classifies as minor error, and the Moon article classifies as not found. Before sharing, I review them though. And IMO these are hallucinations. Specifically built the tool so it is easy to go and review the identified errors and change classification/edit components as you see fit.
1
91
Maybe it is a meta message in the paper itself -- "generating metacognitive feelings of confidence that are easily mistaken for epistemic justification".
14
Hallucination of the day in a Philosophy journal
1
74
Between this (which in my benchmarks appears to work just as well as the typical web search) and Luna, I will likely be able to reduce VerusCite down to $1 to review a paper instead of $2.
Introducing Fast Search in the Perplexity Search API. Fast Search runs on Photon, our new Rust-based retrieval and ranking service that we built with a small team of engineers and hundreds of agents. It returns 95% of search results in 230 ms or less.
90
New blog post, I have been using extra credits to just churn out ideas I have had for years. Peer reviewed papers were always symbolic. The ease you can write them now I am hoping spurs academics to consider "doing things in the real world" as a measure of success.
2
1
4
305
Andy Wheeler retweeted
My feed these days is mostly folks arguing about ALPRs and Flock cameras. In 2025, Omaha launched cityofomahaconnect.org/ I've been curious about it for awhile so I used @CrimeDecoder's new Crime Synth app to estimate monthly violent crime rates after rollout.
2
5
6
1,469
Replying to @KevinARing
I would be happy to put together a proposal if you are open to it. nitter.net/CrimeDecoder/status/21… In addition to the easier macro though, it would be useful to get a breakdown of clearances and when they are useful how it would map to different potential constitutional standards (retrospective search vs real-time exigent, different data retentions, etc.)
San Fran motor vehicle thefts is a good example I like to look at the cumulative differences over time. Any individual month is within the error bar, but it is persistently under predicted. Post June 2024 when they rolled out Flock.
42
Hallucination of the day in BMC Research. Journals could definately afford my service -- they are charging over $3k for APC fees in this journal. Costs only $2 to review in my tool.
1
69
One of the ironies with the Flock losing contracts, would not surprise me if it forced their hand to pivot to the commercial market and make some cheap alternative camera for homes/businesses (competitor to Ring). Which I expected to happen eventually anyway, but may just happen faster now.
1
130
Most recent AI project -- ignorance bands on synthetic control results. So you have many potential solutions. What this does is generate Manski style bands on the feasible solution space (so these are not standard error estimates, but actual treatment effect estimates that are reasonable). It is not as bad as I expected actually.
1
102
San Fran motor vehicle thefts is a good example I like to look at the cumulative differences over time. Any individual month is within the error bar, but it is persistently under predicted. Post June 2024 when they rolled out Flock.
1
2
1,590