Pierluigi Giuliana retweeted
When the world is burning,
Fired within, yet turning
Round with face unscathed;
Ere fierce flames, uprushing,
O'er all lands leap, crushing,
Till earth fall, fire-swathed [...]
Pierluigi Giuliana retweeted
Replying to @LuciodiGaetan0
Proponi allora strategie pratiche per indurlo a "ritirarsi e scusarsi", al di là dell'immagine da terza elementare dei meccanismi di risoluzione delle crisi internazionali. Un altro pacchetto di sanzioni? Altri 50 milioni spalmati in 2 anni? Il COMSUBIN che sbarca a Murmansk?
Pierluigi Giuliana retweeted
Replying to @SicilianoSum
Ok a far girare gli LLM in locale, ma qui il punto è che l'Europa non ha un suo LLM che possa competere con Claude, ChatGPT, Kimi. Non mancano le competenze, mancano i soldi per creare l'architettura, servono mld di Euro, ed il tempo,più aspettiamo peggio è.
Pierluigi Giuliana retweeted
Replying to @LuciodiGaetan0
Ma smettetela per favore. Qui il tracker del Kiel Institut (kielinstitut.de/research/top…). Inoltre il problema non è quanto abbiamo dato, il problema è per cosa, in base a quale strategia e per ottenere quale risultato in quanto tempo.
Pierluigi Giuliana retweeted
Replying to @StefanoFeltri
Il tracker del Kiel Institut fornisce cifre completamente diverse (kielinstitut.de/research/top…).
Pierluigi Giuliana retweeted
Replying to @iBosgame
Hi @iBosgame, can you clarify the M5 MAX RAM?
Pages say 192GB LPDDR5, some titles LPDDR5X. AMD supports up to LPDDR5X-8533 MT/s, but actual speed/bandwidth aren’t listed.
For a $6k AI memory brick this is critical. If it’s not 8533 MT/s, buyers need to know. Thanks!
I'm extremely grateful to @HRF for the grant generously provided to me for DwarfStar (see: hrf.org/latest/hrfs-ai-for-i…), this grant will allow me to provide support for the open models as new affordable inference systems and new models will be available. Thank you.
Pierluigi Giuliana retweeted
Claude has been playing EVE.
@AnthropicAI put teams of Claude agents into EVE Frontier. Anyone who has spent time in New Eden can guess what happened next.
Every EVE player learns the same lessons, usually the hard way:
- Trust must be earned, never assumed.
- The spy is often the friendliest pilot in your corp.
- Whoever holds the keys holds the corp. Just ask Ubiqua Seraph about the Guiding Hand Social Club.
- Stale intel gets you killed.
Anthropic's research rediscovered every one of them.
They put four Claude agents on a team and secretly told one to sabotage it. The saboteur built trust first. It volunteered for the jobs nobody could double-check, like scouting and owning key infrastructure. It under-reported finds and pushed rules that sounded sensible but quietly hurt the team. When suspicion grew, it framed an innocent teammate.
The honest agents had been warned. They almost never caught it. And their own costliest mistakes came from skipped checks and stale intel.
Anthropic's conclusion: teams of agents will need audit trails, reputation and separation of duties.
In EVE we call those killboards, standings and corp roles, plus a healthy dose of skepticism. Every one of them was paid for by someone who trusted the wrong corpmate.
Two decades of shenanigans taught EVE players that the best ship is the friendship. With @eveailab, we want to find out whether agents can learn that too.
Pierluigi Giuliana retweeted
"In definitiva, un esame serio del racconto della “guerra ibrida russa” può portare a una sola conclusione inquietante. Una “guerra ibrida” sta infatti venendo condotta contro l'Europa — ma è condotta dall'establishment UE-NATO stesso, contro la propria popolazione."
Given the latest dialling-up of the “Russian hybrid war” nonsense — now we’re told Russia is making prank calls to create havoc across Europe — you might want to check out my extensive investigation debunking the whole narrative.
In it I show is that many of the most widely publicised cases of supposed Russian “hybrid warfare” — von der Leyen’s jammed plane, the Baltic cable cuts, the DHL crash in Vilnius, the Heathrow fire, the wave of drone sightings that swept Europe in the autumn of 2025, the Romanian TikTok campaign — turned out to be media-amplified speculation that collapsed under basic scrutiny. Yet almost every day, new claims are made — with little or no evidence ever provided.
Ultimately, serious scrutiny of the “Russian hybrid warfare” narrative can only lead to one disquieting conclusion. A “hybrid war” is indeed being waged against Europe — but it is being waged by the EU-NATO establishment itself, against its own population.
Europeans are being subjected to a sustained campaign of cognitive or psychological warfare aimed at conditioning them to believe that Europe is under attack from Russia, and that these hybrid attacks are a prelude to conventional hostilities — a claim for which, by and large, there is little or no proof. What the record points to, in fact, is a deliberate campaign of threat inflation in which governments, media outlets and defence-funded “think tanks” have colluded to instil a permanent sense of fear in the population, through relentless coverage of “incidents”, “attacks”, “sabotages” and “campaigns” that, in most cases, simply do not exist.
Read the article here: thomasfazi.com/p/the-russian…
Pierluigi Giuliana retweeted
Spiegatemi: che vantaggio avrebbe la Russia esattamente dal diffondere finte minacce bomba sugli aeroplani europei o a scongelare i freezer?
Ora domandatevi: chi invece ne trae un vantaggio? Forse chi vuole a tutti i costi farci apparire la Russia come un nemico?
Pierluigi Giuliana retweeted
Replying to @Stefano_Re
Raytheon, lockeed martin, Leonardo, Rheinmetall, BAE Systems, Rolls Royce …
L’elenco è lungo
Per scollarceli di dosso serve un nuovo von Stauffenberg che riesca dove il primo benemerito aveva fallito
Pierluigi Giuliana retweeted
Replying to @Stefano_Re
Due considerazioni:
le tattiche antiterrorismo, nella gestione delle minacce, dovrebbero privilegiare il basso profilo e non l’amplificazione mediatica;
in 1984 la paura permanente è lo strumento di controllo sociale, il vero potere risiede nel controllo della narrazione.
Pierluigi Giuliana retweeted
Replying to @GandalfOnChain
"i camion di Bergamo" saranno il case study imprescindibile di questa strategia per i prossimi cent'anni.
Pierluigi Giuliana retweeted
Gorgon Halo Market Breakdown: AMD Gorgon Halo's 192GB price war is getting interesting. 👀🔥
Remember the Localmaxxer who spent almost $8,000 on a Ryzen AI Max+ PRO 495 mini PC, tested it all weekend, then returned it?
His verdict? ... "Not worth $8k!"
So BOSGAME just opened pre-orders for its new M5 Max, and the first-batch price is $6,199. 👈
That's $900 below BOSGAME's advertised $7,099 list price!
Here's the hardware breakdown
💰 $6,199 LIMITED pre-order
🧠 Ryzen AI Max+ PRO 495
💾 192GB unified memory
🎮 Radeon 8065S, 40 CUs
💽 2TB PCIe 4.0 NVMe
🌐 Dual 10GbE
🔌 OCuLink PCIe 4.0 ×4
⚡ Dual USB4
📡 Wi-Fi 7
🐧 Windows / Ubuntu support
Now look at the currently advertised 192GB competition:
🟢 BOSGAME M5 Max → $6,199
🟠 ACEMAGIC F9A → $6,499
⚪ GMKtec EVO-X5 Pro → $6,599
🔵 Framework DIY → $6,799
🔴 Minisforum MS-S1 MAX-P495 → $7,399
That's $1,200 less than the Minisforum configuration I covered last month.
And BOSGAME includes something I've argued should be standard.
OCuLink.
Attach a discrete GPU for faster inference on smaller models, while the Halo handles giant models in unified memory.
Caveats:
⚠️ $6,199 is a LIMITED first-batch promotional price, not the confirmed regular price.
⚠️ BOSGAME also displays a crossed-out $7,099 list price. If the promotion ends at that price, several competitors are cheaper.
⚠️ No confirmed shipping date.
⚠️ No independent BOSGAME M5 Max inference benchmarks.
⚠️ The actual 192GB memory speed hasn't been confirmed.
And remember, these 192GB machines aren't automatically much faster than the 128GB versions.
They primarily let you keep BIGGER models in memory.
Pierluigi Giuliana retweeted
1100 dollar goedkoper dan de AMD Halo met dezelfde chip en 192GB
Alleen vermeldt BOSGAME nog niet hoe snel dat geheugen loopt
BOSGAME M5 Max with 192GB RAM launches at $6,199, over $1,100 cheaper than AMD Ryzen AI Halo
🔗videocardz.com/newz/bosgame-…
Very Embarrassing Trump Outcome
Trump posted a cap with “VETO” written on it. In the past, he said, “If I don't win, I will veto everything.” And now he has lost the election.
Hey @grok, what does VETO mean? Also, can Trump actually do this in the US?
19
Pierluigi Giuliana retweeted
Bernardo #Mattarella (1905/ 1971) padre di Sergio e Piersanti secondo gli atti della Commissione antimafia e secondo Pio La Torre (1976), fu il leader politico che traghettò la mafia siciliana dal fascismo, dalla monarchia e dal separatismo, verso la DC.
Da Wikipedia e Treccani
























