The Trump administration does not plan to seek approval from Congress for Donald Trump’s planned 250-foot arch,
arguing that they do not need it because lawmakers a century ago authorized a somewhat similar project that was never built.
Trump has targeted Memorial Circle
— a traffic roundabout on Columbia Island, a man-made island tucked inside the edge of Washington
— for his planned arch.
Teams of workers conducting surveys and geophysical testing began work …
good grief i thought bournegol was bad but there are still new depths to plumb
https://www.tuhs.org/cgi-bin/utree.pl?file=V7/usr/src/cmd/sh/mode.h
Ukraine update from Giorgio Provinciali: As a weapons development context, Ukraine offers unrivaled opportunity.
As an engineer, Giorgio distills the marketing speech into reality, and he shares the math.
Share the friend link to bypass the paywall, and help give the world a glimpse of the reality on the ground in Ukraine:
David Louis Hollembaek (Veeva Systems) and Hande Kafkas (cognee) will tackle one of the most pressing AI challenges right now, getting context right for AI agents. Two search veterans, one great session.
Learn more: https://2026.berlinbuzzwords.de/session/search-is-back-solving-the-context-crisis-for-ai-agents/
Get your ticket: https://2026.berlinbuzzwords.de/tickets/
I just had a thought. I have talked to techies who kept their old shop on their resume despite being laid off because "you can't have a gap on your resume." Instead of doing that, just put that you are working for "a stealth-mode startup".
Who are they? I can't tell you. They are in stealth-mode.
Can I get the HR contact to verify employment? I can't. They are in stealth-mode.
What are you doing for them? I can't tell you.![]()
Regret Minimization in Single-Dimensional Contract-Design with Binary Actions
Riccardo Poiani, Martino Bernasconi, Andrea Celli
https://arxiv.org/abs/2606.06125 https://arxiv.org/pdf/2606.06125 https://arxiv.org/html/2606.06125
arXiv:2606.06125v1 Announce Type: new
Abstract: We study principal-agent problems in which a principal commits to an outcome-dependent payment scheme (i.e., a contract) in order to induce an agent to take a costly action leading to a favorable outcome. We consider the online extension of the classical (one-shot) principal-agent problem, in which the principal repeatedly interacts with agents by proposing contracts over multiple rounds. The principal has no information about the agents and, crucially, does not observe their actions. As a result, the principal must learn an optimal contract using only the realized outcomes observed at each round. We focus on the setting with binary actions and single-dimensional agent types, where the agent's private type represents their cost per unit-of-effort. For adversarial-type sequences, we provide tight $\Theta(T^{2/3})$ regret guarantees. Remarkably, this rate is completely independent of the number of outcomes $m$. The upper bound is based on two key components: 1) a reduction to a one-dimensional threshold optimization problem and 2) a non-uniform discretization to handle the non-Lipschitz nature of the problem. Moreover, in the case of a single (fixed) hidden type, we show that it is possible to improve the rates and provide a tight $\widetilde{\Theta}(\sqrt{T})$ regret bound. Our algorithm is based on an explore-then-commit strategy where we first approximately learn the hidden type via a stochastic binary search, and then we commit to a ``robustified'' near-optimal contract.
toXiv_bot_toot
It’s Primary Eve in New York.
On the ballot on Tuesday?
A series of raucous Democratic contests with the potential to radically reshape New York City’s congressional delegation and push it further to the left.
Plus, a Democratic race to challenge Representative Mike Lawler,
a vulnerable Republican incumbent who holds a Hudson Valley-area seat that Democrats are desperate to flip.
But here we will focus largely on the New York City primaries,
several of wh…
Too Good to Be True: A Study on Modern Automatic Speech Recognition for the Evaluation of Speech Enhancement
Danilo de Oliveira, Tal Peer, Timo Gerkmann
https://arxiv.org/abs/2605.12107 https://arxiv.org/pdf/2605.12107 https://arxiv.org/html/2605.12107
arXiv:2605.12107v1 Announce Type: new
Abstract: Speech enhancement (SE) systems are typically evaluated using a variety of instrumental metrics. The use of automatic speech recognition (ASR) systems to evaluate SE performance is common in literature, usually in terms of word error rate (WER). However, WER scores depend heavily on the choice of ASR system and text normalization pipeline. In this paper, we investigate how modern ASR models correlate with human recognition of enhanced speech. A listening experiment reveals that modern ASR models with large-scale noisy training and embedded language models correlate more with human WER than simpler ones, with a transducer model providing the most reliable transcriptions. Nevertheless, we also show that these models' robustness to noise and use of context can be uninformative to an acoustics-focused evaluation of enhancement performance.
toXiv_bot_toot
In the cramped, dilapidated Ebola ward, a 5-year-old boy languished on a bare mattress, a tissue stuffed into his nose to stanch the incessant bleeding. His father stood over him, eyes clouded with worry.
A few beds away lay the body of Christiane Bahati, 21, who had died seven hours earlier but had not yet been taken away. Her shoes were still tucked under the bed, her wailing relatives gathered outside the ward doors.
The body, covered by a thin sheet, was highly contagious. Ye…