We need tactical retrieval possums, someone get on this
November 8, 2026 at 3:08 PM
We need tactical retrieval possums, someone get on this
[System Message: Wolffie-unit has reached a critical point of overstimulation. Core meltdown imminent. Sanity retrieval maneuvers are offline at this time. Please evacuate for your own safety.]
November 10, 2026 at 12:41 AM
[System Message: Wolffie-unit has reached a critical point of overstimulation. Core meltdown imminent. Sanity retrieval maneuvers are offline at this time. Please evacuate for your own safety.]
It's systems thinking, not systems changing.
That's a different department.
That's a different department.
November 9, 2026 at 3:04 AM
It's systems thinking, not systems changing.
That's a different department.
That's a different department.
I integrate digital tools into daily bell-ringer routines and formative retrieval practice—using Kahoot! for quick checks and Google Classroom
November 7, 2026 at 3:08 PM
I integrate digital tools into daily bell-ringer routines and formative retrieval practice—using Kahoot! for quick checks and Google Classroom
Are current LLMs better at simple one-shot factual queries? Yes, very much so. Can you rely on them for deep research and long context retrieval/analysis? No, and they're not usable at all without multiple layers of verification and adjudication, where they still have 20-50% unsupported claim rates.
November 8, 2026 at 3:47 PM
Are current LLMs better at simple one-shot factual queries? Yes, very much so. Can you rely on them for deep research and long context retrieval/analysis? No, and they're not usable at all without multiple layers of verification and adjudication, where they still have 20-50% unsupported claim rates.
Meta-analysis of retrieval practice, spacing, and interleaving in classrooms concludes fairly robust average effects of each (open access) link.springer.com/article/10.1...
November 6, 2026 at 10:52 AM
Meta-analysis of retrieval practice, spacing, and interleaving in classrooms concludes fairly robust average effects of each (open access) link.springer.com/article/10.1...
Ainda tem pouco estudo sobre por ser um fenômeno recente, mas tem esse ensaio aqui por exemplo
Separaram grupos d pessoas entre as q assistiam videos curtos e as que viam vídeos longos
O cérebro do grupo d videos curtos funcionava de forma PIOR q o outro, com piora da memória e dificuldade cognitiva
Separaram grupos d pessoas entre as q assistiam videos curtos e as que viam vídeos longos
O cérebro do grupo d videos curtos funcionava de forma PIOR q o outro, com piora da memória e dificuldade cognitiva
November 10, 2026 at 2:26 AM
Ainda tem pouco estudo sobre por ser um fenômeno recente, mas tem esse ensaio aqui por exemplo
Separaram grupos d pessoas entre as q assistiam videos curtos e as que viam vídeos longos
O cérebro do grupo d videos curtos funcionava de forma PIOR q o outro, com piora da memória e dificuldade cognitiva
Separaram grupos d pessoas entre as q assistiam videos curtos e as que viam vídeos longos
O cérebro do grupo d videos curtos funcionava de forma PIOR q o outro, com piora da memória e dificuldade cognitiva
me being very interested in info retrieval as a concept and how information is disseminated across the web: :D
me taking the info retrieval midterm:
me taking the info retrieval midterm:
November 6, 2026 at 10:25 PM
me being very interested in info retrieval as a concept and how information is disseminated across the web: :D
me taking the info retrieval midterm:
me taking the info retrieval midterm:
In my info retrieval class, a professor—an old school Google designer way before “go ahead, be evil” was the company line—showed us all these systems designed in the 90s-00s. One search used the human tendency to remember fragments of information. There was one that was legit story-oriented.
My boss went on a rant about how browser bookmarks and tabs are two different ways of doing the same thing, and now I can't stop thinking about it
But I dared to post this on Mastodon first and the replies were "they are not the same at ALL, you FOOL"
No capacity to entertain a novel perspective!
But I dared to post this on Mastodon first and the replies were "they are not the same at ALL, you FOOL"
No capacity to entertain a novel perspective!
November 7, 2026 at 4:24 PM
In my info retrieval class, a professor—an old school Google designer way before “go ahead, be evil” was the company line—showed us all these systems designed in the 90s-00s. One search used the human tendency to remember fragments of information. There was one that was legit story-oriented.
MIT researchers revealed CacheGraft, enhancing retrieval-augmented generation using key-value cache translation. It achieved improved answer quality in 29 of 30 tests, without deploying the larger model, promising benefits in resource-constrained contexts. https://arxiv.org/abs/2611.07393
CacheGraft: Enhancing Retrieval-Augmented Generation via Latent-Space Document Enrichment
ArXiv link for CacheGraft: Enhancing Retrieval-Augmented Generation via Latent-Space Document Enrichment
arxiv.org
November 7, 2026 at 7:20 PM
MIT researchers revealed CacheGraft, enhancing retrieval-augmented generation using key-value cache translation. It achieved improved answer quality in 29 of 30 tests, without deploying the larger model, promising benefits in resource-constrained contexts. https://arxiv.org/abs/2611.07393
The reason the oldest skyscrapers are in coastal cities is so that the elevators could safely land in the ocean for retrieval
November 6, 2026 at 4:59 PM
The reason the oldest skyscrapers are in coastal cities is so that the elevators could safely land in the ocean for retrieval
Sasuke Retrieval Arc went hard as fuck
November 8, 2026 at 8:46 PM
Sasuke Retrieval Arc went hard as fuck
RAG steht für Retrieval-Augmented Generation. Im Prinzip erklären die Guides wie man Prompts formulieren sollte, damit ein Modell nicht nur auf seine Trainingsdaten zugreift.
November 8, 2026 at 9:19 AM
RAG steht für Retrieval-Augmented Generation. Im Prinzip erklären die Guides wie man Prompts formulieren sollte, damit ein Modell nicht nur auf seine Trainingsdaten zugreift.
Subtle deficits in word retrieval, processing speed, or executive function may become apparent only during demanding work.
I would expect a full recovery, but moments of weakness visible that may preclude a leadership challenge without knowing more.
I would expect a full recovery, but moments of weakness visible that may preclude a leadership challenge without knowing more.
November 7, 2026 at 1:38 PM
Subtle deficits in word retrieval, processing speed, or executive function may become apparent only during demanding work.
I would expect a full recovery, but moments of weakness visible that may preclude a leadership challenge without knowing more.
I would expect a full recovery, but moments of weakness visible that may preclude a leadership challenge without knowing more.
First up, a paper I loved from @yamilrvelez.bsky.social and Alec Ewig on ways to use AI within web surveys that makes the surveys easier and more relevant for respondents (which not all uses of AI do), identifying specific policy topics from open ends and creating tailored conjoints
Retrieval-Augmented Surveys | Political Analysis | Cambridge Core
Retrieval-Augmented Surveys
www.cambridge.org
November 6, 2026 at 2:39 PM
First up, a paper I loved from @yamilrvelez.bsky.social and Alec Ewig on ways to use AI within web surveys that makes the surveys easier and more relevant for respondents (which not all uses of AI do), identifying specific policy topics from open ends and creating tailored conjoints
Ich helfe grundsätzlich gerne, aber manchmal wundert es mich schon, warum mich Leute beispielsweise fragen, was RAG ist, statt einfach kurz in die Wikipedia oder eine andere zuverlässige Quelle zu schauen. Ich könnte schließlich auch den größten Unsinn erzählen.
de.wikipedia.org/wiki/Retriev...
de.wikipedia.org/wiki/Retriev...
Retrieval-Augmented Generation – Wikipedia
de.wikipedia.org
November 8, 2026 at 9:24 AM
Ich helfe grundsätzlich gerne, aber manchmal wundert es mich schon, warum mich Leute beispielsweise fragen, was RAG ist, statt einfach kurz in die Wikipedia oder eine andere zuverlässige Quelle zu schauen. Ich könnte schließlich auch den größten Unsinn erzählen.
de.wikipedia.org/wiki/Retriev...
de.wikipedia.org/wiki/Retriev...
And would have a very poor data retrieval system!
November 8, 2026 at 9:16 AM
And would have a very poor data retrieval system!
Scaling Laws For Data-Constrained Dense Retrieval
Studies how the best attainable dense retrieval loss scales with model size and unique fine-tuning data when data is reused. Finds a multiplicative power law across BERT-family retrievers on MS MARCO.
📝 arxiv.org/abs/2611.04979
Studies how the best attainable dense retrieval loss scales with model size and unique fine-tuning data when data is reused. Finds a multiplicative power law across BERT-family retrievers on MS MARCO.
📝 arxiv.org/abs/2611.04979
Scaling Laws For Data-Constrained Dense Retrieval
Dense retrieval models are typically obtained by fine-tuning pretrained language models on supervised retrieval data, where high-quality annotations or interaction logs are often limited and reused ac...
arxiv.org
November 4, 2026 at 9:27 PM
Scaling Laws For Data-Constrained Dense Retrieval
Studies how the best attainable dense retrieval loss scales with model size and unique fine-tuning data when data is reused. Finds a multiplicative power law across BERT-family retrievers on MS MARCO.
📝 arxiv.org/abs/2611.04979
Studies how the best attainable dense retrieval loss scales with model size and unique fine-tuning data when data is reused. Finds a multiplicative power law across BERT-family retrievers on MS MARCO.
📝 arxiv.org/abs/2611.04979
You might be able to get away with a fairly straightforward local setup that keeps you away from any complicated storage/retrieval besides backing up a SQLite file
November 9, 2026 at 4:18 PM
You might be able to get away with a fairly straightforward local setup that keeps you away from any complicated storage/retrieval besides backing up a SQLite file
Persistent Retrieval for Conversational Agents: A Study of Retrieval Efficiency
Karthik Ravula et al.
#arXiv #cs.AI #cs.CL #cs.IR
Karthik Ravula et al.
#arXiv #cs.AI #cs.CL #cs.IR
Persistent Retrieval for Conversational Agents: A Study of Retrieval Efficiency
Conversational agents can retrieve information directly or delegate retrieval to a subagent. This design choice affects how retrieval history is maintained, what context each model call receives, and what information must be re-retrieved in follow-up questions. We compare three designs: skill-based…
arxiv.org
November 6, 2026 at 10:11 PM
Yes, but only if we also get to name the retrieval mechanism Pole Daniels
November 5, 2026 at 8:34 AM
Yes, but only if we also get to name the retrieval mechanism Pole Daniels
AI content retrieval 'already eating away' at publisher businesses, court told - via @pressgazette.co.uk & @charlottetobitt.bsky.social #seo
AI content retrieval 'already eating away' at publisher businesses, court told
Unique visitors to leading US online publishers fell by a quarter between 2023 and 2025, according to new data from Digital Content Next.
pressgazette.co.uk
November 9, 2026 at 5:40 AM
AI content retrieval 'already eating away' at publisher businesses, court told - via @pressgazette.co.uk & @charlottetobitt.bsky.social #seo
Sometimes it makes me wonder if anyone teaches information retrieval in CS classes anymore. Did it degrade into ERMs and relational database design?
November 8, 2026 at 2:02 PM
Sometimes it makes me wonder if anyone teaches information retrieval in CS classes anymore. Did it degrade into ERMs and relational database design?
For a reliable RAG system, track source freshness, retrieval quality, answer grounding and whether users can inspect citations. Fluency alone is not a quality metric. #RAG #EnterpriseAI
November 9, 2026 at 9:55 AM
For a reliable RAG system, track source freshness, retrieval quality, answer grounding and whether users can inspect citations. Fluency alone is not a quality metric. #RAG #EnterpriseAI