Ukrainian-founded language learning marketplace Preply raised a $150M Series D led by WestCap at a $1.2B valuation; the startup has a 150-person office in Kyiv (Anna Heim/TechCrunch)
https://techcrunch.com/2026/01/21/langu…
Trump Needs a Shrink and a Baby-Sitter, Not a National Security Adviser (emptywheel)
https://www.emptywheel.net/2026/01/20/trump-needs-a-shrink-and-a-baby-sitter-not-a-national-security-adviser/
http://www.memeorandum.com/260120/p68#a260120p68
from my link log —
Banish: a declarative DSL embedded in Rust, for defining rule-based state machines.
https://github.com/LoganFlaherty/banish
saved 2026-02-20 https:…
Anthropic rolls out embedded security scanning for Claude
https://cyberscoop.com/anthropic-claude-code-security-automated-security-review/
Exploiting ID-Text Complementarity via Ensembling for Sequential Recommendation
Liam Collins, Bhuvesh Kumar, Clark Mingxuan Ju, Tong Zhao, Donald Loveland, Leonardo Neves, Neil Shah
https://arxiv.org/abs/2512.17820 https://arxiv.org/pdf/2512.17820 https://arxiv.org/html/2512.17820
arXiv:2512.17820v1 Announce Type: new
Abstract: Modern Sequential Recommendation (SR) models commonly utilize modality features to represent items, motivated in large part by recent advancements in language and vision modeling. To do so, several works completely replace ID embeddings with modality embeddings, claiming that modality embeddings render ID embeddings unnecessary because they can match or even exceed ID embedding performance. On the other hand, many works jointly utilize ID and modality features, but posit that complex fusion strategies, such as multi-stage training and/or intricate alignment architectures, are necessary for this joint utilization. However, underlying both these lines of work is a lack of understanding of the complementarity of ID and modality features. In this work, we address this gap by studying the complementarity of ID- and text-based SR models. We show that these models do learn complementary signals, meaning that either should provide performance gain when used properly alongside the other. Motivated by this, we propose a new SR method that preserves ID-text complementarity through independent model training, then harnesses it through a simple ensembling strategy. Despite this method's simplicity, we show it outperforms several competitive SR baselines, implying that both ID and text features are necessary to achieve state-of-the-art SR performance but complex fusion architectures are not.
toXiv_bot_toot
The God in the Wire: The Book That Began with an Empty Shelf
I did not set out to write a book about technology. I set out to understand an empty shelf. The shelf is at LaGuardia Community College in Long Island City, Queens, mounted on a corridor wall beneath a sign bearing the universal symbol for Deaf access. The shelf once held a TTY, one of those text telephones that gave Deaf people their first access to instantaneous distance communication.
Cardinals, Raiders, Vikings, Rams named top landing spots for embattled Pro Bowl quarterback https://www.sportingnews.com/us/nfl/miami-dolphins/news/cardinals-raiders-vikings-rams…
OpenAI and ServiceNow sign a three-year deal to integrate OpenAI's models into ServiceNow's business software, including embedding OpenAI's AI agents (Belle Lin/Wall Street Journal)
https://www.wsj.com…