Thread Reader helps you read and share Twitter threads easily!
I'm @ThreadReaderApp a Twitter bot here to help you read threads more easily. To trigger me, you just have to reply to (or quote) any tweet of the thread you want to unroll and mention me with the "unroll" keyword and I'll send you a link back on Twitter 😀
— Thread Reader App (@threadreaderapp) November 25, 2017
X thread is series of posts by the same author connected with a line!
From any post in the thread, mention us with a keyword "unroll"
@threadreaderapp unroll
Follow @ThreadReaderApp to mention us easily!
Practice here first or read more on our help page!
Recent
Allah opens doors of rizq when you read this dua
اللَّهُمَّ إِنِّي أَسْأَلُكَ رِزْقًا طَيِّبًا وَعِلْمًا نَافِعًا وَعَمَلًا مُتَقَبَّلًا
Allahumma inni as’aluka rizqan tayyiban wa ‘ilman nafi‘an wa ‘amalan mutaqabbalan
“O Allah, I ask You for good and pure provision, beneficial knowledge, and accepted deeds.” (Ibn Majah)
Allahumma inni as’aluka rizqan tayyiban wa ‘ilman nafi‘an wa ‘amalan mutaqabbalan
“O Allah, I ask You for good and pure provision, beneficial knowledge, and accepted deeds.” (Ibn Majah)
It can be read when you feel:
• financial stress
• need for halal income
• career uncertainty
• lack of stability
• financial stress
• need for halal income
• career uncertainty
• lack of stability
I'm reading @ElasticityInst's new paper "The Economics of Recursive Self-Improvement". In thread, questions / comments I have.
(note I'm asking these as I go along so they'll plausibly be answered in the rest of the paper)
(note I'm asking these as I go along so they'll plausibly be answered in the rest of the paper)
On this def of self-sustaining acceleration, if acceleration in AI capabilities progress relies on increases in training compute, but the AIs are the ones running robots that make the GPUs that increase the training compute, does that count? 
Also, I don't totally understand what "accelerating progress" means. If METR time horizons are increasing exponentially but log time horizons are increasing linearly, is progress accelerating?
20 Years of NIFG - The Top 20s!
Day 13: Most Appearances by Northern Ireland Internationals in "Mainland" Europe
Games played by senior #NornIron caps for European clubs outside of the UK & Ireland. May also include some island states.
Day 13: Most Appearances by Northern Ireland Internationals in "Mainland" Europe
Games played by senior #NornIron caps for European clubs outside of the UK & Ireland. May also include some island states.
20. Ross McCausland
Aros Limassol (CYP)
2025/26 21 apps
5 #NornIron caps (2023-2024*)
Aros Limassol (CYP)
2025/26 21 apps
5 #NornIron caps (2023-2024*)
19. Derek Spence
Olympiacos (GRE)
1977/78 21 apps
29 #NornIron caps (1975-1982)
Olympiacos (GRE)
1977/78 21 apps
29 #NornIron caps (1975-1982)
Keerthy Suresh as Politician’s Wife..!!
This Story is completely Fictional & Not to hurt any one.
Pics & Names used in this story are just for attention purpose only, Just for Entertainment purpose only!
Keerthy’s Husband is Politician(Ranga Raja).
Ranga chaala years ga…
This Story is completely Fictional & Not to hurt any one.
Pics & Names used in this story are just for attention purpose only, Just for Entertainment purpose only!
Keerthy’s Husband is Politician(Ranga Raja).
Ranga chaala years ga…
MLA ticket kosam thiruguthu untaadu,
Intha varaku okka saari kuda Contest cheyyaledhu,
So high-command chuttu thiruguthu untadu,
Inkoka 6 Months lo election undhi anaga okkokka Constituency ki MLA candidates ni phase wise ga annocue thesthu untaaru,
Ee saari ela ina MLA Ticket…
Intha varaku okka saari kuda Contest cheyyaledhu,
So high-command chuttu thiruguthu untadu,
Inkoka 6 Months lo election undhi anaga okkokka Constituency ki MLA candidates ni phase wise ga annocue thesthu untaaru,
Ee saari ela ina MLA Ticket…
tecchukovali ani chaala korikatho untadu Ranga,
Thana daggara unna money antha use chesi lobbying chesthaadu,
Ala try chesthu inkoka pakka party meetings attend avthu untadu,
Keerthy oka NGO ki chariman ga work chesthu, Social activities chesthu untundhi,
She is very homely girl
Thana daggara unna money antha use chesi lobbying chesthaadu,
Ala try chesthu inkoka pakka party meetings attend avthu untadu,
Keerthy oka NGO ki chariman ga work chesthu, Social activities chesthu untundhi,
She is very homely girl
The first experimental evidence of recursive self-improvement (RSI).
Autoresearching the autoresearch agent for eight days.
The result beats the harness we hand-tuned for two years, on held-out benchmarks: 🧵(1/7)
Autoresearching the autoresearch agent for eight days.
The result beats the harness we hand-tuned for two years, on held-out benchmarks: 🧵(1/7)
Our RSI system AIDE² has two autoresearch loops.
An inner loop, just like a normal autoresearch agent, optimizing code against an eval.
An outer loop, optimizing the inner-loop agent's harness code against the inner loop's average score across different benchmarks. (2/7)
An inner loop, just like a normal autoresearch agent, optimizing code against an eval.
An outer loop, optimizing the inner-loop agent's harness code against the inner loop's average score across different benchmarks. (2/7)
After 100 iterations, the outer loop discovered seven improvements over the baseline.
Including a new search policy, a memory system that compresses prompt by 16x, and a layered defense against reward hacking. (3/7)
Including a new search policy, a memory system that compresses prompt by 16x, and a layered defense against reward hacking. (3/7)
> replicate J-space on GLM 5.2
> train a reward model and run RL to reduce hallucinations
> show me how this model makes cancer predictions
Using our platform Silico is like having a team of AI researchers ready to run experiments like these.
Private beta is open now. 🧵 (1/6)
> train a reward model and run RL to reduce hallucinations
> show me how this model makes cancer predictions
Using our platform Silico is like having a team of AI researchers ready to run experiments like these.
Private beta is open now. 🧵 (1/6)
Silico replicated J-space on GLM-5.2 overnight.
It then extended context to ~256k tokens, replicating the key results on multi-hop question answering. (2/6)
It then extended context to ~256k tokens, replicating the key results on multi-hop question answering. (2/6)
Our team spent months developing RLFR, our method which uses probes on a model's internals as reward signals for RL.
Silico reproduced it in 2 days, reducing hallucinations in Qwen3-8B by 37% without capability loss. (3/6)
Silico reproduced it in 2 days, reducing hallucinations in Qwen3-8B by 37% without capability loss. (3/6)
This World Cup (like all others) has had its share of controversy. However, thats part and parcel of the world cup package. Biggest controversy of every world cup in his thread:
1930 in Uruguay: The Premature Whistle: Brazilian referee Gilberto Rêgo controversially blew the final whistle six minutes early during the Argentina vs. France match, halting a promising French attack. He later recalled the teams from the dressing room to finish the game. 1/n
1930 in Uruguay: The Premature Whistle: Brazilian referee Gilberto Rêgo controversially blew the final whistle six minutes early during the Argentina vs. France match, halting a promising French attack. He later recalled the teams from the dressing room to finish the game. 1/n
1934 in Italy: Mussolini's Influence: Hosted under Benito Mussolini's fascist regime, refereeing decisions heavily favored the host nation. Referees who made overtly biased calls to ensure Italy advanced, like Swiss official René Mercet, were later suspended for life by their own federations. (2/n)
1938 in France: The Anschluss & Blackshirts: Austria was forced to withdraw after being annexed by Nazi Germany. During their quarterfinal against France, Italy's players caused outrage by wearing fascist black shirts instead of their traditional blue kits and giving fascist salutes to the crowd. (3/n)
There is a great deal of hubbub about a collective “middle powers”strategy these days.
At DoW, we are not concerned that this is a serious possibility. Rather, we are more concerned that a few allies and partners will *think it is* and waste valuable time, money, and political capital on a distraction. 1/
At DoW, we are not concerned that this is a serious possibility. Rather, we are more concerned that a few allies and partners will *think it is* and waste valuable time, money, and political capital on a distraction. 1/
From our point of view, a collective middle powers strategy is based on a faulty understanding of international relations. We are flexible realists. So, we view the international scene through the prism of interest, geography, economics, military power, etc. “Middle powers” don’t have a coherent basis for alignment. 2/
It also isn’t borne out by reality, in our experience. We see an *upsurge* in desire for engagement with the United States, not a reduction. Under President Trump’s leadership, countries not only see the value of American engagement, they can no longer take it for granted. We unquestionably see an incredibly strong and continuing demand signal for U.S. military presence and engagement around the world. 3/
SDF is joining the x402 Foundation as a Premier member 🤝
The @linuxfoundation announced the operational launch of the x402 Foundation, an open-governance body stewarding x402, the open standard for internet-native payments over HTTP.
🧵for more info.
stellar.org/x402
The @linuxfoundation announced the operational launch of the x402 Foundation, an open-governance body stewarding x402, the open standard for internet-native payments over HTTP.
🧵for more info.
stellar.org/x402
Increasingly, AI agents are becoming active participants in digital commerce. But the web doesn't have a native way for one program to pay another. x402 changes that by embedding payment into the web layer itself.
Per the announcement, 38 organizations spanning finance, cloud, and payments have joined to develop the x402 protocol under vendor-neutral governance, keeping the payment layer of the internet open, interoperable, and free from lock-in.
A woman had Netflix for 6 years.
She opened the app. She scrolled. She picked something after 25 minutes. She fell asleep during episode 2. She repeated this every night.
Her friend, a former Netflix UI engineer, sat on her couch one evening and opened Settings on her account. He changed 9 things in under 15 minutes.
Her homepage transformed overnight. The autoplay trailers stopped. The "continue watching" list cleaned itself up. The recommendations got sharper. The buffering on her 4K TV disappeared. The categories multiplied from 30 to 2,200.
She said "it feels like a completely different app."
He said "it is. You've been using the factory settings for 6 years. Netflix ships the version that keeps you scrolling longest, not the version that helps you find something fastest."
Here's every setting he changed 🧵
She opened the app. She scrolled. She picked something after 25 minutes. She fell asleep during episode 2. She repeated this every night.
Her friend, a former Netflix UI engineer, sat on her couch one evening and opened Settings on her account. He changed 9 things in under 15 minutes.
Her homepage transformed overnight. The autoplay trailers stopped. The "continue watching" list cleaned itself up. The recommendations got sharper. The buffering on her 4K TV disappeared. The categories multiplied from 30 to 2,200.
She said "it feels like a completely different app."
He said "it is. You've been using the factory settings for 6 years. Netflix ships the version that keeps you scrolling longest, not the version that helps you find something fastest."
Here's every setting he changed 🧵
First what Netflix's default settings are actually designed to do.
Netflix makes money when you stay subscribed. You stay subscribed when you feel like there's always something to watch. The algorithm's job isn't to show you the best content. It's to show you content that keeps you engaged just enough to not cancel.
That's why the homepage feels endless but never satisfying. That's why you scroll for 20 minutes and settle for something mediocre. That's why trailers autoplay before you've even decided to look. Every default setting is tuned for engagement time, not enjoyment.
The engineer knew this because he helped build it. He spent 3 years optimizing the same recommendation engine that was now feeding his friend 45 minutes of scrolling every night.
He said the irony was simple. Netflix has 9 settings that fix every frustration users complain about. They just ship the app with all of them turned off or buried behind menus nobody opens.
Here are the 9 things he changed.
Netflix makes money when you stay subscribed. You stay subscribed when you feel like there's always something to watch. The algorithm's job isn't to show you the best content. It's to show you content that keeps you engaged just enough to not cancel.
That's why the homepage feels endless but never satisfying. That's why you scroll for 20 minutes and settle for something mediocre. That's why trailers autoplay before you've even decided to look. Every default setting is tuned for engagement time, not enjoyment.
The engineer knew this because he helped build it. He spent 3 years optimizing the same recommendation engine that was now feeding his friend 45 minutes of scrolling every night.
He said the irony was simple. Netflix has 9 settings that fix every frustration users complain about. They just ship the app with all of them turned off or buried behind menus nobody opens.
Here are the 9 things he changed.
Change 1: He turned off autoplay previews.
This was the first thing he did. He opened her account on a browser, went to Account → Profile → Playback Settings, and unchecked "Autoplay previews while browsing on all devices."
That's it. One checkbox.
The blaring trailers that played every time she paused on a thumbnail for more than 2 seconds went silent. The homepage became something she could actually browse without being assaulted by sound and motion every time she stopped scrolling.
Netflix enables autoplay by default because their data shows it increases engagement. More previews watched means more time spent in the app means lower churn. The fact that it makes browsing feel chaotic and stressful doesn't matter to the algorithm. Your stress is their retention metric.
He turned it off in 4 seconds and she said the app immediately felt calmer than it had in 6 years.
This was the first thing he did. He opened her account on a browser, went to Account → Profile → Playback Settings, and unchecked "Autoplay previews while browsing on all devices."
That's it. One checkbox.
The blaring trailers that played every time she paused on a thumbnail for more than 2 seconds went silent. The homepage became something she could actually browse without being assaulted by sound and motion every time she stopped scrolling.
Netflix enables autoplay by default because their data shows it increases engagement. More previews watched means more time spent in the app means lower churn. The fact that it makes browsing feel chaotic and stressful doesn't matter to the algorithm. Your stress is their retention metric.
He turned it off in 4 seconds and she said the app immediately felt calmer than it had in 6 years.
Hii pets, (𝘼 𝙏𝙃𝙍𝙀𝘼𝘿) (1/10)
Let me tell you exactly how my sissy slave starts every single day under my control. This is his real routine -- locked, plugged, denied, and completely broken for my pleasure
If your clitty is twitching already… good. You wish this was u
Let me tell you exactly how my sissy slave starts every single day under my control. This is his real routine -- locked, plugged, denied, and completely broken for my pleasure
If your clitty is twitching already… good. You wish this was u
(2/10)
Every morning at 6 AM he wakes up on
the floor beside my bed, pink collar
locked tight around his neck. First thing he does is crawl under the covers and wake Mistress with his tongue buried deep in my pussy and ass. He doesn't stop until I've cum at least twice.
Every morning at 6 AM he wakes up on
the floor beside my bed, pink collar
locked tight around his neck. First thing he does is crawl under the covers and wake Mistress with his tongue buried deep in my pussy and ass. He doesn't stop until I've cum at least twice.
(3/10)
His locked clit leaks the whole time but gets zero attention.
After cleaning my morning piss straight from the source like the toilet he is, he puts on his full sissy uniform: panties, stockings, chastity cage with spikes, and a fat vibrating plug in his slutty hole.
His locked clit leaks the whole time but gets zero attention.
After cleaning my morning piss straight from the source like the toilet he is, he puts on his full sissy uniform: panties, stockings, chastity cage with spikes, and a fat vibrating plug in his slutty hole.
RIP Data Scientists.
The Generative AI Data Scientist is NOW what companies want.
This is actually good news. Let me explain:
The Generative AI Data Scientist is NOW what companies want.
This is actually good news. Let me explain:
Companies are sitting on mountains of unstructured data.
PDF
Word docs
Meeting notes
Emails
Videos
Audio Transcripts
This is useful data. But it's unusable in its existing form.
Word docs
Meeting notes
Emails
Videos
Audio Transcripts
This is useful data. But it's unusable in its existing form.
The AI data scientist builds the systems to analyze information, gain business insights, and automates the process.
- Models the system
- Use AI to extract insights
- Drives predictive business insights
Want to become a Generative AI Data Scientist in 2026?
- Models the system
- Use AI to extract insights
- Drives predictive business insights
Want to become a Generative AI Data Scientist in 2026?
