berbie

Not just tackling bouldering problems #BigHairyGoal #TrailRunner #MacAppStore #iOSAppStore #bhag

[berbie @berbie]

Brainstorming and progress tracking for creatives and creators with the right balance of structure and freedom.

[berbie @berbie]

@theepicmap Adam Ondra

[berbie @berbie]

@burkov It's all about explaining what you really want. The 80% are whatever everyone wants, and that is in the corpus already. Then you need to define real convergence criteria to get the rest. And if you are not up to the task mentally, the AI can't help you either.

[berbie @berbie]

@burkov Prepare for /compact early is my favorite on this. Then ask what’s missing from the plan and continue next step

[Epic Maps 🗺️ @theepicmap]

The Most Dangerous Animals in the World.

[David @david_eng_mba]

The most profound human insights often emerge from the intersection of disciplines, where diverse perspectives collide and new ideas are born. This insight highlights the power of blending knowledge from different fields—science, art, technology, or philosophy—to spark innovation and understanding. Rather than seeing disciplinary boundaries as limits, it views them as opportunities for synergy. When seemingly unrelated ideas meet, they create a space for unexpected connections and solutions, revealing truths that might remain hidden within a single domain. This perspective encourages us to embrace complexity and diversity in thought, recognizing that the edges of knowledge are where breakthroughs thrive.

[berbie @berbie]

@burkov @ChapterPal Is ChapterPal staff a bunch of LLMs?

[berbie @berbie]

@Giovann35084111 Let me know when I can become one of the first members!

[berbie @berbie]

@Giovann35084111 We are deviating around the PowerLaw line. In either case if you take the area above the line and subtract it from the area below the line it is a zero sum. And if there are no highe bubbles, then the lower area also becomes less wide and shorter. As what we've seen so far.

[BURKOV @burkov]

In this remarkable paper, a team at DeepMind and Terence Tao addressed a fundamental challenge in mathematics using an AI system called AlphaEvolve. The challenge is the following:

Many practical math problems ask "what's the best possible arrangement?"—like packing the most circles in a square or finding the largest set of points with no isosceles triangles—and answering these requires actually finding specific examples from search spaces so large that systematically exploring them by hand is impractical.

AlphaEvolve doesn't have the LLM directly generate candidate solutions (which would require millions of expensive LLM calls to explore the search space)—instead, it evolves Python programs that search for solutions, so one expensive LLM call generates a search heuristic that then runs cheaply for millions of iterations exploring candidate configurations, and the population of these search programs evolves over time with better ones surviving, essentially doing meta-optimization where the thing being optimized is itself an optimizer.

Across 67 problems spanning combinatorics, geometry, and analysis, AlphaEvolve discovered new constructions improving decades-old results (hexagon packing, Heilbronn triangles, sum-product problems) while matching state-of-the-art in dozens of others.

The paper reports every problem they tried including failures where AlphaEvolve couldn't match known bounds, which makes this the most thorough empirical evaluation of AI for mathematical discovery.

Read and ask questions on ChapterPal: https://t.co/WGlSXoJesU

Download the PDF: https://t.co/bLOX7ve9GD

[Alison Fisk @AlisonFisk]

Something beautiful for the weekend!

An extraordinary Acheulean handaxe knapped around a fossil shell around 500,000-300,000 years ago!

The shell’s central display has been described as an example of early aesthetic intention by the toolmaker.

From West Tofts, Norfolk.

📷 Museum of Archaeology and Anthropology, University of Cambridge https://t.co/8psldM5ZQy

#Archaeology

[berbie @berbie]

@moneyordebt Hi Steven! Thank you for your great research! Could you publish your current parameter set on the period since 2023?

[berbie @berbie]

@WalrusProtocol Urslaw

[Sabine Hossenfelder @skdh]

as of the time of writing, the only way to create consciousness is having children

[berbie @berbie]

DeekSeek’s approach is like the shift from 5GHz Intel Pentiums to 1GHz multicore IBM PowerPCs: smaller specialized models, non-overlapping knowledge, smart caches/pipelines to combine results. Likely trains models sequentially on the same hardware—saves resources, not time.

[berbie @berbie]

@Giovann35084111 @mememememeemmee Thank you Gio for everything you discover and share. I highly appreciate your appetite on knowledge and insight. I am now a patreon supporter although to just understand the cyclical and PowerLaw bound nature of bitcoin is what everybody can learn here on X for free. Keep on!

[berbie @berbie]

@dotkrueger @bensig Thanks for the suggestions! I would say either choose 50% controversial or go straight and add
@Giovann35084111
over
@BritishHodl

[berbie @berbie]

@TheRobynHD War ja abzusehen. Da ist sie, die Reichensteuer. Eigentlich ist das eine gute Sache. Bin mal gespannt, was das Bundesverfassungsgericht dazu sagt. Dann müsste man Oldtimer, Rolex und den Weinkeller auch besteuern.

[berbie @berbie]

@Sina_21st @xan_nick @PeterZeihan Hm, to me this reads like his argument no longer applies anyway, as economic growth is no longer in sync with the money supply.

[berbie @berbie]

@Giovann35084111 Gio! Thank you for your work and your contribution! You remind me of my strict Latin teacher, who insisted uncompromisingly on accuracy, and at the same time, you are open to new ideas and let them bestow fresh insights.

[berbie @berbie]

@RaoulGMI @Kevin_Kelly_II Thank you for this linguistic inspiration. I just consulted the trusted AI of my choice to explore the differences between hyperstition, narrative, and meme—and was thoroughly educated.