You mean the JFK documents that were released containing hundreds of social security numbers of federal employees? Not exactly a glowing recommendation.

AI scrapers use these residential proxies. It's not (just) VPNs and Tor routing. Several bottom-feeding companies openly advertise such scraping services, for pretty much any country you may want. I administer a wiki that's been on the receiving end of such scraping, and the majority of these scraping requests are in fact coming from residential IP-addresses rather than data centers.

I don't know whether these are hacked accounts, people getting tricked or paid to run these scraping apps on their devices, but it's impossible to block them all. Even if you let fail2ban block entire /24s for every detected hit (even disregarding the collateral damage and the fact that these blocks don't solve the issue, the fail2ban and iptables overhead starts to outweigh the apache load at some point).

Anubis seems to be taking care of it for now, but it's obviously only a matter of time before they can deal with that one too. Although its delay does enable fail2ban rules to block the IP-addresses before they get to stress the mediawiki php scripts, attempting to diff 2 revisions of a random page from 10 years ago.

For the Irish language course the recordings of native speakers were taken offline in 2023. The AI replacements are nonsensical.

This story is about AI generated courses, not voices, but my post was still (accidentally) on-topic: when they previously used AI to increase volume of content, they were ok with quality being thrown out the window.

The AI generated courses might be low quality, and the original (English) courses might also go downhill because the type of exercises they produce may now be restricted to the type of things that their AI is able to reorganise for other languages. E.g. it might go further in the direction of vocabulary memorisation.

Isn't it that they call it "español" in Latin America, and "castellano" in Spain?

Unfortunately, the voices are really bad.

It's a pity they don't also make available the old courses, with audio from native speakers.

Maybe from NOAA? (at least until they get defunded)

Wow, you really can't even continue reading once sentence after the one that triggers you.

Everything you said is true for an individual, but misses the point the parent was making regarding the total number of charging stations needed.

If an EV needs to charge 20 minutes every 200 miles, and an ICE vehicle needs to pump gas for 5 minutes every 200 miles, then you will need 4 times the number of public level 3 chargers as gas pumps to serve the same number of cars. In fact if roadtripping EVs are mostly charging at normal meal breaks, then you will need even more chargers, since use will be more clustered compared to gas stops.

On the flip side, the public chargers will only be used by people on road trips or who don't have a home charger which will decrease their volume. Still I would expect you to need more chargers than gas pumps along highways, and fewer in the middle of cities.

They wrote this software for free, and they gave you for free security updates that don't require accepting any new features. They even went out of their way to continue putting out security releases (ESR 115.13) for operating systems that Microsoft and Apple don't even support anymore. And they did this all six months in advance so you would have plenty of time to upgrade.

But enjoy your freedom to run old unpatched software.

In many ways I like the Lyttle Lytton contest even more. There were some very good Bulwer-Lytton entries, but it could also devolve into who could write the longest run-on sentence. Having a short word limit forced authors to really distill their horribleness.

If you take a book, reproduce it exactly, and add commentary at the end, the resulting work is a derivative work and you need permission of the original copyright holder to distribute it. FSF has only granted permission to use the AGPL exactly as it is with no modification or additions, so Neo4j had no permission to use their modified version of the AGPL, and doing so was violating FSF's copyright.

FSF's Zoe Kooyman and Krzysztof Siewicz will give a presentation on Sunday 2nd of Feb:

"FSF's criteria for free machine learning applications"

https://fosdem.org/2025/schedu...

It'll be streamed. Well worth tuning in for. A recording should be online soon after.

ELIZA's evil twin.

The Fed wants to keep the details secret and vary them frequently so the banks aren't over optimizing to the test. This makes senset, for example with standardized crash testing it is common for cars to perform great on the exact crash scenario tested for, but poorly on off-nominal crash scenarios that occur in the real world.

On the other hand, secret laws and regulations are antithesis to free society because how can you justly expect anyone to follow rules when you won't tell them what the rules are. And since the banks are required to modify their behavior if they fail these tests, then the tests are effectively regulations.

A fair solution would be to have the test published ahead of time, but still change frequently adapting to what the Fed perceives as gaming of the test. The problem is that going through the normal regulation process with comment periods and such takes a good two years at least, and the Fed would like to move faster than that.

I really wanted cryptocurrency to work out. There is such a need for digital cash that lets people pay for things online without credit card companies charging exorbitant processing fees and then on top of that tracking every transaction and selling your purchase history to anyone willing pay. In addition to making existing purchases better, I thought cryptocurrency had a chance of becoming Clay Shirky's micropayments, enabling business models that can't exist now. Reading about far away lands (SV) where there were actually bitcoin ATMs in bars, it sounded like the future was almost on us.

Instead it turned out worse than credit cards or even paypal in every way. Your purchase history isn't anonymous. It is at best pseudonymous but easily traceable if you use it as widely as you you would a credit card, and thus effectively public unless you go to great lengths to obscure it. Processing fees are higher than credit cards. And it requires insane resources to operate that would make the most inefficient monopolist bureaucracy blush in embarrassment. It is completely useless as a normal day-to-day currency.

Then to make matters worse, it attracted a huge number of grifters, conmen, speculators, and criminals who are its only real users.

Etherium is at least trying to work towards being useful as a day-to-day payment method. It has been slow progress, and I don't know if they will ever get there, but at least there is some progress. Bitcoin though is failed experiment. It had it's run, and it is time to put it down.