Plates can have variable number of characters, which makes it nonsensical to define OCR accuracy per plate. So, OCR accuracy of 96% must be per character. Flock have not defined what they mean by OCR accuracy as far as I can find. In such case the standard per-character approach to calculating accuracy applies.

Yes I know, this is Slashdot and you are supposed to comment without reading the article. Well I did read the article and wouldn't you know it:

The Flock cameras were not the only variable that changed. By the looks of it it isn't even the most significant variable. Nothing in the articles speaks of what each of the variables contributed to the final 11% reduction. All of the evidence is quotes of people with vested interest saying "Trust me bro, it is the cameras". Well in that case all I can say is "Trust me bro, it is the cops, the bait cars, and the aggressive prosecution"

According to flock's own press releases their OCR accuracy is 96%. This means that the accuracy of reading a six character plate number is ~78%. This is likely reflecting the best case scenario and not real life where lighting, dirt, motion, weird angles, and obstacles are all degrading the accuracy.

So besides the couple of criminals that they caught, can we please hear about all the people who got stopped, potentially arrested and perhaps even had guns drawn on them? Also are they saying that they would have never caught the criminals by other means if it wasn't for the Flock cameras.

Don't get me wrong, I see the value of these cameras for solving crime. I think this is undeniable. What is also undeniable is the misuse and the deleterious effect mass surveillance have on democracy. Unless there is an independent and transparent oversight system that ensures this data is used for the cime purposes being claimed I think it does more bad than good. "Trust me bro" is not going to cut it for me. What is also important is who is physically in possession of the primary data and its derivatives, and how is their use of the data ensured to be lawful. It doesn't seem that the local law enforcement olds the data.

I watched the videos. I am going to pull the wifi card from the TV and have it connected only over ethernet. Luckily, where the TV is located it can't reach the neighbours open Xfinity AP.

I also have some words: "Optional my a*s!" Two weeks ago my LG TV sdid software update. After that it would not let me watch anything until I agreed to all terms, including the ones about collecting and sharing of all kinds of data including fingerprinting what is being played over HDMI from say a laptop. So I either have to agree to being spied on or be unable to use the TV a paid for. I agreed to the terms and then went on to block any known server it pings back to. Since then I am monitoring the firewall logs to make sure it is not phoning a server I don't know of. Anyone know of a TV that is not a snooping device?

Ah, the priors strike again! There should be a mandate that every accuracy claim to comes with the expected positive predictive value and the number of expected false positives.

He does not need to liquidate. He can borrow, tax free, against his paper assets - that's what everyone in his position has been doing for years. To repay the loan, he just gets another one. A trillion dollars in paper assets can go very, very far especially if you allocate a small fraction for bribes to keep the rules of the game benefiting you. Things like buying yourself a government or two.

I imagined it and the market didn't last long - all the school kids were wiped out within a month of school starting.

Shut up Ivan, you are not a UK citizen.

I am sure saving 43 minutes a day as determined by a Linked-in post will be a great benefit to everyone's health. I hope they checked the accuracy of the AI output. Because it is not like UK will ever prosecute and convict innocent employees based on the inaccurate output of a computer system. That, would never happen, right?

Does grossed out count as entertained?

Shhhh... you are giving away the grift.

Two a*holes were planing to screw each other and take control over openAI. One got ahead while the other one was distracted. Now they sue each other over who was supposed to take it over.

There is a special place in hell specifically for people who extrapolate from limited data. People who do that from meta-data are the worst. So let's not put to much weight on projections based on effects within the error margin of a random collection of studies.

Considering that the politicians you are talking about are sponsored by the people the tax is aimed at, this proposition may kill two birds with one stone.