Republished under Creative Commons, Original article from Pluralistic.

Well, fuck.

Last year, Google lost an antitrust case to Biden's DoJ. The DoJ lawyers beat Google like a drum, proving beyond a shadow of a doubt that Google had deliberately sought to create and maintain a monopoly over search, and that they'd used that monopoly to make search materially worse, while locking competitors out of the market.

In other words, the company that controls 90% of search attained that control by illegal means, and, having thus illegitimately become the first port of call for the information-seeking world, had deliberately worsened its product to make more money:

pluralistic.net/2024/04/24/nam…

That Google lost that case was a minor miracle. First, because for 40 years, the richest, most terrible people in the world have been running a literal re-education camp for judges where they get luxe rooms and fancy meals and lectures about how monopolies are good, actually:

pluralistic.net/2021/08/13/pos…

But second, because Judge Amit Mehta decided that the Google case should be shrouded in mystery, suppressing the publication of key exhibits and banning phones, cameras and laptops from the courtroom, with the effect that virtually no one even noticed that the most important antitrust case in tech history, a genuine trial of the century, was underway:

promarket.org/2023/10/27/googl…

This is really important. The government doesn't have to win an antitrust trial in order to create competition. As the saying goes, "the process is the punishment." Bill Gates was so personally humiliated by his catastrophic performance at his deposition for the Microsoft antitrust trial that he elected not to force-choke the nascent Google, lest he be put back in the deposition chair:

pluralistic.net/2020/09/12/wha…
a
But Judge Mehta turned his courtroom into a Star Chamber, a black hole whence no embarrassing information about Google's wicked deeds could emerge. That meant that the only punishment Google would have to bear from this trial would come after the government won its case, when the judge decided on a punishment (the term of art is "remedy") for Google.

Yesterday, he handed down that remedy and it is as bad as it could be. In fact, it is likely the worst possible remedy for this case:

gizmodo.com/google-wont-have-t…

Let's start with what's not in this remedy. Google will not be forced to sell off any of its divisions – not Chrome, not Android. Despite the fact that the judge found that Google's vertical integration with the world's dominant mobile operating system and browser were a key factor in its monopolization, Mehta decided to leave the Google octopus with all its limbs intact:

pluralistic.net/2024/11/19/bre…

Google won't be forced to offer users a "choice screen" when they set up their Android accounts, to give browsers other than Chrome a fair shake:

pluralistic.net/2024/08/12/def…

Nor will Google be prevented from bribing competitors to stay out of the search market. One of the facts established in the verdict was that Google had been slipping Apple more than $20b/year in exchange for which, Apple forbore from making a competing search engine. This exposed every Safari and iOS user to Google surveillance, while insulating Google from the threat of an Apple competitor.

And then there's Google's data. Google is the world's most prolific surveiller, and the company boasts to investors about the advantage that its 24/7 spying confers on it in the search market, because Google knows so much about us and can therefore tailor our results. Even if this is true – a big if – it's nevertheless a fucking nightmare. Google has stolen every fact about our lives, in service to propping up a monopoly that lets it steal our money, too. Any remedy worth the name would have required Google to delete ("disgorge," in law-speak) all that data:

pluralistic.net/2024/08/07/rev…

Some people in the antitrust world didn't see it that way. Out of a misguided kind of privacy nihilism, they called for Google to be forced to share the data it stole from us, so that potential competitors could tune their search tools on the monopolist's population-scale privacy violations.

And that is what the court has ordered.

As punishment for being convinced of obtaining and maintaining a monopoly, Google will be forced to share sensitive data with lots of other search engines. This will not secure competition for search, but it will certainly democratize human rights violations at scale.

Doubtless there will be loopholes in this data-sharing order. Google will have the right to hold back some of its data (that is, our data) if it is deemed "sensitive." This isn't so much a loophole as is a loopchasm. I'll bet you a testicle⹋ that Google will slap a "sensitive" label on any data that might be the least bit useful to its competitors.

⹋not one of mine

This means that even if you like data-sharing as a remedy, you won't actually get the benefit you were hoping for. Instead, Google competitors will spend the next decade in court, fighting to get Google to comply with this order.

That's the main reason that we force monopolists to break up after they lose antitrust cases. We could put a bunch of conditions on how they operate, but figuring out whether they're adhering to those conditions and punishing them when they don't is expensive, labor-intensive and time consuming. This data-sharing wheeze is easy to do malicious compliance for, and hard to enforce. It is not an "administrable" policy:

locusmag.com/2022/03/cory-doct…

This is all downside. If Google complies with the order, it will constitute a privacy breach on a scale never before seen. If they don't comply with the order, it will starve competitors of the one tiny drop of hope that Judge Mehta squeezed out of his pen. It's a catastrophe. An utter, total catastrophe. It has zero redeeming qualities. Hope you like enshittification, folks, because Judge Mehta just handed Google an eternal licence to enshittify the entire fucking internet.

It's impossible to overstate how fucking terrible Mehta's reasoning in this decision is. The Economic Liberties project calls it "judicial cowardice" and compared the ruling to "finding someone guilty for bank robbery and then sentencing him to write a thank you note":

economicliberties.us/press-rel…

Matt Stoller says it's typical of today's "lawlessness, incoherence and deference to big business":

thebignewsletter.com/p/a-judge…

David Dayen's scorching analysis in The American Prospect calls it "embarassing":

prospect.org/justice/2025-09-0…

Dayen points out the many ways in which Mehta ignored his own findings, ignored the Supreme Court. Mehta wrote:

This court, however, need not decide this issue, because there are independent reasons that remedies designed to eliminate the defendant’s monopoly—i.e., structural remedies—are inappropriate in this case.

Which, as Dayen points out is literally a federal judge deciding to ignore the law "because reasons."

Dayen says that he doesn't see why Google would even bother appealing this ruling: "since it won on almost every point." But the DoJ could appeal. If MAGA's promises about holding Big Tech to account mean anything at all, the DoJ would appeal.

I'll bet you a testicle⹋ that the DoJ will not appeal. After all, Trump's DoJ now has a cash register at the reception desk, and if you write a check for a million bucks to some random MAGA influencer, they can make all charges disappear:

pluralistic.net/2025/09/02/act…

⹋again, not one of mine

And if you're waiting for Europe to jump in and act where America won't, don't hold your breath. EU Commission sources leaked to Reuters that the EU is going to drop its multi-billion euro fine against Google because they don't want to make Trump angry:

reuters.com/legal/litigation/g…

Sundar Pichai gave $1m to Donald Trump and got a seat on the dais at the inaguration. Trump just paid him back, 40,000 times over. Trump is a sadist, a facist, and a rapist – and he's also a remarkably cheap date.

• Proofpoint researchers observed an increase in opportunistic cybercriminals using malware based on Stealerium, an open-source malware that is available “for educational purposes.”
• Multiple other stealers share significant code overlap with Stealerium, such as Phantom Stealer. Throughout this blog post, we’ll use the name Stealerium to refer to infostealers that share significant overlap with the original Stealerium.
• Threat actors are increasingly pivoting to information stealers, as targeting identity becomes a priority for cybercriminals.

Not Safe for Work: Tracking and Investigating Stealerium and Phantom Infostealers | Proofpoint US

Key findings  Proofpoint researchers observed an increase in opportunistic cybercriminals using malware based on Stealerium, an open-source malware that is available “for educational

^Proofpoint

::: spoiler Comments
- Hacker News.
:::

• 50% of young men say they would rather date an AI girlfriend than risk rejection from a human partner
• 31% of U.S. men aged 18–30 report already chatting with AI girlfriends
• 19% of American adults overall say they have explored AI romance
• 80% of Gen Z say they would consider a virtual relationship with an AI girlfriend
• 83% of Gen Z believe they can form a deep emotional bond with AI companions

Half of Young Men Would Rather Date an AI Girlfriend Than Face Loneliness or Rejection, New Report Reveals

AI girlfriends go mainstream: 80% of Gen Z open to virtual relationships as AI girlfriend market nears $140 billion

^{Girlfriend.ai (Half of Young Men Would Rather Date an AI Girlfriend Than Face Loneliness or Rejection, New Report Reveals)}

Despite having pledged to reach net-zero emissions, major members of Net Zero Asset Managers hold billions of dollars’ worth of fossil-fuel stocks, including those in “carbon bomb” projects, while marketing their funds as green and sustainable.

Meet the UN-backed 'green' investors’ group that invested in fossil fuels - DeSmog

Despite having pledged to reach net-zero emissions, major members of Net Zero Asset Managers hold billions of dollars' worth of fossil-fuel stocks, including those in “carbon bomb” projects, while marketing their funds as green and sustainable.

^{Stefano Valentino (DeSmog)}

• noyb.

noyb WIN: French DPA fines Google €325 million for “Spam Emails” in Gmail

The CNIL fined Google €325 million for creating spam emails in Gmail

^noyb.eu

Sondaggione per il libro di settembre e di conseguenza quello di ottobre

[poll type="regular" results="on_close" public="true" chartType="bar" close="2025-09-06T22:00:00.000Z" groups="BookClub"]

• Le Cosmicomiche - Italo Calvino (proposto da @levysoft)
• Le venti giornate di Torino - Giorgio De Maria (proposto da @yaku)
• Il grande ritratto - Dino Buzzati (proposto da @levysoft)
• I reietti dell’altro pianeta - Ursula K. Le Guin (proposto da @yaku)

[/poll]

Scadenza sondaggio il 07/09 alle 00:00 così lunedì si parte!

Discutine sul nostro forum.

Sondaggio libri Settembre-Ottobre 2025

Sondaggione per il libro di settembre e di conseguenza quello di ottobre sondaggio Scadenza sondaggio il 07/09 alle 00:00 così lunedì si parte! 🙂

^{Gatti Ninja}

https://www.kickstarter.com/projects/doctorow/enshittification-the-drm-free-audiobook

Recently tried an Impossible burger and nuggets and thought that if nobody told me it wasn't meat, I'd have thought the patty was made out of a weird kind of meat, rather than make a connection with the taste and texture of plants. Honestly, I might not complain if that was the only kind of "meat" I could have for the rest of my life.

Well, maybe I'd miss bacon.

I've yet to find the opportunity to try lab-grown meat, but I for sure would like to try it out and don't see much wrong with it as long as it's sustainable, reasonably priced, and doesn't have anything you wouldn't expect in a normal piece of meat.

Also, with imitation and lab-grown options, I'd no longer have to deal with the disgust factor of handling raw meat (esp. the juices) or biting into gristle. I'll happily devour a hot dog, but something about an unexpected bit of cartilage gives me a lingering sense of revulsion.

https://www.bloomberg.com/news/articles/2025-09-03/australia-backs-coal-mine-extension-stoking-climate-backlash?accessToken=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzb3VyY2UiOiJTdWJzY3JpYmVyR2lmdGVkQXJ0aWNsZSIsImlhdCI6MTc1Njg4MDYxNSwiZXhwIjoxNzU3NDg1NDE1LCJhcnRpY2xlSWQiOiJUMVpVNkFHUDlWREMwMCIsImJjb25uZWN0SWQiOiIwRDVEOUMzOUI5REE0MDhGQTJEODdFQTYzOEEyNzU4MSJ9._fJXGSng4LjJc9DJhAUoaqaHyTiACIk_Au_OOAOBtSM&leadSource=uverify+wall

France's data protection authority on Wednesday issued record fines against search giant Google and fast-fashion platform Shein for failing to respect the law on internet cookies.

The two groups, each with tens of millions of users in France, received two of the heaviest penalties ever imposed by the CNIL watchdog: 150 million euros ($175 million) for Shein and 325 million euros for Google.

Cookies are small files saved to browsers by websites that can collect data about users' online activity, making them essential to online advertising and the business models of many large platforms.

France fines Google, Shein record sums over cookie law violations

France’s data protection watchdog CNIL on Wednesday fined Google €325 million ($380 million) and fast-fashion retailer Shein €150 million ($175 million) for violating cookie rules.

^{FRANCE 24}

cross-posted from: retrolemmy.com/post/24567714

A new way to collect music that matters.

Bandcamp Clubs

A Bandcamp-style, subscribe-to-own music discovery experience, curated by trusted experts and rooted in community.

^bandcamp.com

More Golden Girls action:

TAS s1e15 "Eye of the Beholder"

Cory Doctorow is rightfully enraged:

This is all downside. If Google complies with the order, it will constitute a privacy breach on a scale never before seen. If they don't comply with the order, it will starve competitors of the one tiny drop of hope that Judge Mehta squeezed out of his pen. It's a catastrophe. An utter, total catastrophe. It has zero redeeming qualities. Hope you like enshittification, folks, because Judge Mehta just handed Google an eternal licence to enshittify the entire fucking internet.

Jeffrey Epstein’s accusers went to the Capitol to ask Congress to get behind their calls for more disclosures, but momentum for a bill demanding it appeared to stall.

https://www.nytimes.com/2025/09/03/us/politics/epstein-bill-republicans-trump.html?unlocked_article_code=1.jE8.9ThW.suzq7W1tOdQ9

The Hawthorne Caballeros performing “On The Edge” during the 2025 DCI All-Age Championship Finals at Lucas Oil Stadium in Indianapolis, Indiana.

All of these photos are available under a Creative Commons license, free for you to use as long as you give me photography credit.
Hawthorne Caballeros
2025 DCI All-Age Championship Finals
Photo Credit: Kevin Gamin
You can find all of the edited photos from this and other events on my Flickr site.
Hawthorne Caballeros
2025 DCI All-Age Championship Finals
Photo Credit: Kevin Gamin
You can find all of my photos on my Smugmug site.
Hawthorne Caballeros
2025 DCI All-Age Championship Finals
Photo Credit: Kevin Gamin

cross-posted from: tucson.social/post/2212969

A 32-year-old man Mexican man died of unknown causes on Sunday after being detained by U.S. Immigration and Customs Enforcement at a private prison in Arizona, authorities confirmed.

Mexican man dies in immigration detention in Arizona - tucson.social

Lemmy

^{tucson.social}

Visti gli imprevisti con HaxeFlixel che non ho ancora avuto il tempo di elaborare qui, stavo (ri)considerando il basato Love2D che, ultimamente mi sono (ri)accorta, gira su talmente tante piattaforme da rendere inutile anche fare degli esempi qui. La cosa seccante di quel coso, però, è che non è esattamente un motore di gioco, quanto […]

octospacc.altervista.org/2025/…

minioctt

2025-09-03 22:43:56

l’amore 2d e gli oggetti orientifici, ma quello che accade non fa divertire (test e valutazioni prestazioni OOP su Love2D)

Visti gli imprevisti con HaxeFlixel che non ho ancora avuto il tempo di elaborare qui, stavo (ri)considerando il basato Love2D che, ultimamente mi sono (ri)accorta, gira su talmente tante piattaforme da rendere inutile anche fare degli esempi qui. La cosa seccante di quel coso, però, è che non è esattamente un motore di gioco, quanto più un framework multimediale… e quindi, a differenza di Flixel e altri robini, non ha tutte le varie utilità che è bene avere per poter sviluppare qualcosa senza partire dallo zero assoluto… e quindi, l’idea sarebbe di creare una specie di motorino per esso per gestire cose comuni come sprite, fisica, boh, queste cose (non) belle. 🤥

Ovviamente, il problema inevitabile è sorto immediatamente dopo una giornata di lavoro iniziale — a dire il vero, fatta per fortuna con le pinze, perché ero sotto sotto pronta a vedere cose storte accadere — e cioè che, con una dose di OOP in realtà nemmeno troppo grossa per gli standard comuni, le prestazioni sono crollate così tanto a picco che, per un semplice giochino di Breakout (la demo di HaxeFlixel, che ho adattato strada facendo per testare), da un lato su PC l’uso di CPU si aggirava attorno al 10-15% (che è tipo il quintuplo di cosa fa la stessa demo in HaxeFlixel)… mentre, su piattaforme pazzurde come il 3DS era letteralmente ingiocabile, facendo 5 secondi a frame (e pensare che io ho il new, che è più veloce). 💀

Ho fatto un po’ di ricerca e — per quanto fosse a me comunque ovvio che una programmazione ad oggetti basata principalmente sull’ereditarietà rende un programma più lento, perché i sassi elettronici sono fatti per eseguire istruzioni sequenziali e lavorare con memoria quanto più continua possibile, che è il contrario di cosa succede con tutte quelle classi che estendono classi che estendono il mercato che mio padre comprò — non immaginavo che su Lua il calo prestazionale fosse tale da essere non solo evidente, ma proprio fuori scala in certi casi. E ora, dunque, i problemi sono grossissimi. 😤

Anche stavolta ho raccolto molti link a riguardo di questa ennesima causa di sofferenza per me, e in realtà ancora non ho capito bene la questione, ma un grande problema sembra essere causato dagli accessi a tabelle nidificate, e alle chiamate di funzioni fatte più del necessario anche per operazioni altrimenti veloci… e nel mio caso certamente una buona parte di overhead in questo senso sarà causata dal fatto di non scrivere Lua nativo, bensì usare Haxe (o in alternativa, TypescriptToLua) per traspilare a Lua, ma sentivo che il problema non poteva essere solo il codice bloattato generato da questi affari… 🧨

E infatti, scrivendo in Lua puro un piccolo benchmark (battezzato al volo solo per dare un titolo al memo: “Love2D fucking rectangles“, genera innumerevoli rettangoli e li fa muovere calcolando le collisioni), prima in modo classico e poi con un minimo di OOP, ed eseguendolo oltre i limiti del ragionevole, ho visto le cose brutte: la versione OOP è in effetti più lenta. Non tanto più lenta, e comunque dipende dalle opzioni con cui la si fa girare, ma solo perché è comunque molto semplice… a differenza del motorino che tanto vorrei creare per replicare la API di HaxeFlixel in Love2D per quanto possibile (evidentemente, non molto possibile). 😭

Dopo ben 4 (quattro) immagini non so se ho voglia di elaborare oltre… Ma, in sostanza: in una modalità, il programma genera solo X (200mila) rettangoli all’avvio, mentre nell’altra ne genera X (200) a frame, andando all’infinito, calcolando sempre le collisioni… e quindi con la prima si esclude una lentezza dovuta alla continua istanziazione di oggetti, mentre la seconda da modo di vedere come un programma rallenta nel tempo rispetto all’altro (generando meno oggetti a parità di tempo). 💥

Nella prima modalità, il carico è basato principalmente sull’accesso alla chiamata draw, quindi non potevo limitare il numero di quadrati effettivamente visibili, e quindi ho potuto eseguirla solo su PC, dove si nota in media un rallentamento di circa il doppio per la versione OOP, che accede a svariate proprietà nidificate per fare il disegno… Mentre, nella seconda la prova il carico era più il resto, quindi ho deciso di limitare il numero di rettangoli visibili a schermo ad ogni frame agli ultimi X (500), e questo mi ha permesso di eseguire il programma pure sul 3DS senza che crashasse (credo ci siano limiti di VRAM lì), ma sia su PC che su 3DS si vede che la versione non-OOP riesce a generare in media 1,2 sprite in più per delta di tempo, differenza che nel corso di minuti diventa di migliaia di sprite. 😵

Incredibile, spassoso, magicante, ma… e adesso??? Boh! Dovrò ingegnarmi pesantemente per creare un motorino sufficientemente generalizzato da poter essere usato come comoda libreria per molti giochi Love2D, ma che allo stesso tempo sia efficiente… ma qui casca l’asino, perché per implementare concetti come uno sprite, che oltre ai classici dati come posizione X e Y ha un oggetto “disegnabile”, che può essere un’immagine o una forma geometrica, che quindi richiede chiamate della API Love2D completamente diverse dietro le quinte, non vedo alternativa non incasinata se non l’OOP; ma non basterà usare più la composizione che l’ereditarietà, bensì per sconfiggere l’overhead serviranno mosse di design interne talmente scomode che ho davvero tanta paura anche solo a pensare di scriverle… 😱

#benchmark #development #LOVE2D #Lua #test #testing

Memo by ██▓▒­░⡷⠂𝚘𝚌𝚝𝚝 𝚒𝚗𝚜𝚒𝚍𝚎 𝚞𝚛 𝚠𝚊𝚕𝚕𝚜⠐⢾░▒▓██

TypeScriptToLua, A generic TypeScript to Lua transpiler: + https://typescripttolua.github.io + https://github.com/TypeScriptToLua/TypeScriptToLua 

^Memos