Christian Neisler
4 followers · 6 following · 663 views
on the atlas — 47
- When AI builds itself \ Anthropic34 savers
- Understand - a novelette by Ted Chiang6 savers
- Reward hacking is becoming more sophisticated and deliberate in frontier LLMs — LessWrong2 savers
- Predicting our own demise | Reducible Errors2 savers
- long degeneracy - by jez - oldcoin bad, newcoin good2 savers
- The 24 Hour Life - peepeepoopoo1 savers
- Robert Wiblin - about me, links to writing, contact1 savers
- A potential employee’s guide to Silicon Valley startup equity3 savers
- Dario Amodei — Machines of Loving Grace34 savers
- Book Summary: Zero to One — LessWrong2 savers
- How To Become A Mechanistic Interpretability Researcher — LessWrong2 savers
- Clawed - by Dean W. Ball - Hyperdimensional2 savers
- escaping flatland: career advice for CS undergrads43 savers
- About Us3 savers
- The Multi-Armed Bandit Problem and Its Solutions | Lil'Log3 savers
- Dario Amodei — The Adolescence of Technology44 savers
- Cache Lines - Algorithmica2 savers
- Thoughts on AI 2027 - Machine Intelligence Research Institute2 savers
- weekly dinners in sf3 savers
- Sub-70 half marathon training using a percentage-based approach - Running Writings1 savers
- leogao's Shortform — LessWrong1 savers
- Nicholas Fabiano, MD on X: "Cortisol from exercising suppresses the next cortisol response to a stressor. Exercise before doing something stressful. https://t.co/FnxgD5Atl5" / X1 savers
- KV Caching Explained: Optimizing Transformer Inference Efficiency3 savers
- Checking in on AI-2027 — LessWrong1 savers
- The cumulative damage theory of injuries - Running Writings1 savers
- Wejo Speaks: Why I Sucked in College - LetsRun.com1 savers
- Goal Factoring - LessWrong8 savers
- CFAR Handbook — LessWrong2 savers
- Why Were Early Psychedelicists So Weird? | Slate Star Codex1 savers
- Do One New Thing A Day To Solve Your Problems — LessWrong1 savers
- What are your greatest one-shot life improvements? - LessWrong4 savers
- Pause Your Feedback Loops2 savers
- Curius / Onboarding2621 savers
- Curius / Bookmarks for the extremely curious182 savers
- The Bitter Lesson78 savers
- HOWTO: Be more productive (Aaron Swartz's Raw Thought)34 savers
- Impact, agency, and taste | benkuhn.net30 savers
- How do you find your people, when searching is itself an antipattern? | Kevin Liu20 savers
- Aman's AI Journal • Primers • Ilya Sutskever's Top 3020 savers
- Six Selections on Self-Defense - by Andrew Wu11 savers
- Being the (Pareto) Best in the World - LessWrong11 savers
- Some Not Entirely Mainstream Tips for Meeting a Partner10 savers
- What 2026 looks like - LessWrong8 savers
- Here’s how I use LLMs to help me write code3 savers
- Sandbagging with misaligned action — Chain-of-Thought Transcript — Anti-Scheming3 savers
- A (Long) Brain Dump on Investment Banking Recruiting3 savers
- Why business majors play with crayons - by Justin Guo2 savers
highlights — 68
if Unversal Basic Income is ever implemented, it will accelerate long degeneracy to new levels
long degeneracy - by jez - oldcoin bad, newcoin goodThree stages: Learn the ropes (≤1 month) learn the essentials, go breadth-first; Learn with research mini-projects practice basic research skills with 1-5 day mini projects, focus on fast feedback loop skills; Work up to full projects, do 1-2 week research sprints, continue the best ones. Explore deeper skills and the mindset of a great researcher.
How To Become A Mechanistic Interpretability Researcher — LessWrongIf we do live in that world, on the other hand, the Trump Administration just cast itself as the enemy of the industry that is about to birth the most powerful technology ever conceived—as well as an enemy of the technology itself.
Clawed - by Dean W. Ball - Hyperdimensionalif you are having an amazing time meeting lots of new cool people, then other people will be more excited to meet you and willing to help you on your quest;
leogao's Shortform — LessWronga huge part of having a good social life is just taking social bids whenever they become available.
leogao's Shortform — LessWrongI prefer 8 or 9 hours between runs if I'm doubling.
The cumulative damage theory of injuries - Running WritingsSo, it's better to do 3km-2km-2km-1km-1km-1km than the other way around.
The cumulative damage theory of injuries - Running WritingsDoing a long run only every two weeks is advantageous because it allows you to do 14-day training cycles and avoid the race-long run-workout triple that is the bane of many college runners. You should realize that your long run will likely be the most damaging workout you do and you should treat your recovery as such. Younger runners would do well with an off-day after a true long run, and more experienced runners would do well to take an easy day.
The cumulative damage theory of injuries - Running Writingslong run 1.5x further than your typical main run of the day
The cumulative damage theory of injuries - Running WritingsSo practical experience tells us that twice a day the max for reasonable people.
The cumulative damage theory of injuries - Running Writingsstress over the duration of a run is cumulative
The cumulative damage theory of injuries - Running WritingsI went out and ran the best race of my life a 28:06 pr to get 4th at nationals.
Wejo Speaks: Why I Sucked in College - LetsRun.comNow, a bad race, a bad workout doesn’t faze me. I know running fast is not that complicated.
Wejo Speaks: Why I Sucked in College - LetsRun.comI always said that if I didn’t get nervous before races I would quit running because what would be the point of competing
Wejo Speaks: Why I Sucked in College - LetsRun.comMost college runners I know don’t seem to know how to do a tempo run properly, they either run it too hard like a race or too easy like an uptempo jog. You want to be approaching your threshold. This could be the topic of another training piece.
Wejo Speaks: Why I Sucked in College - LetsRun.comIf you’re one of these guys, the temptation is to get back to school and start pounding away at your intervals once you start working out. Just try to resist the urge to do this
Wejo Speaks: Why I Sucked in College - LetsRun.comFor almost every workout, you should be finishing faster than you start, and you should be trying to find a relaxed rhythm not trying to necessarily run the set of intervals as fast as possible on that day.
Wejo Speaks: Why I Sucked in College - LetsRun.comsearching for the “secret” to success instead of just getting out and getting in consistent aerobic training.
Wejo Speaks: Why I Sucked in College - LetsRun.comIf I had stayed focused on the bigger picture I should have run a lot faster in college. I don’t think I was destined to be a college star, but my success after college and some flashes of promise that I showed in college suggested that I should have raced faster while in school. As I said distance running is primarily an aerobic activity so it makes sense that I or anyone else for that matter should continue to improve once they get out of school. My improvement was much better than most.
Wejo Speaks: Why I Sucked in College - LetsRun.comI was often running hard intervals or hard fartlek but rarely got in the pace work that I feel can really help build your endurance. In college, I might be doing intervals at 65 pace, easy runs at 6:30, and the rare tempo work I did might be in a workout at 6:00-5:40 pace for most of it.
Wejo Speaks: Why I Sucked in College - LetsRun.comObviously you would like to run on as soft a surface as possible but there have been people who have run high mileage on pavement.
Wejo Speaks: Why I Sucked in College - LetsRun.comI told him to slow down. Sure enough he called me back a few weeks later and was amazed that he had done his first 70 mile week.
Wejo Speaks: Why I Sucked in College - LetsRun.comLooking back I think I would have been a lot better off if I took a longer term approach and tried to consistently get in as many miles as possible and not worry so much about the short term.
Wejo Speaks: Why I Sucked in College - LetsRun.comSo you need to know the purpose of each workout, each recovery run, and take it to heart. Most young guys think they just need to run “harder” to be good. Often this is not the case. They just need to run faster. Think “fast” not “hard.” I ran plenty hard in college, and some of that was my problem.
Wejo Speaks: Why I Sucked in College - LetsRun.comIf I’m trying to bang out every fartlek, every interval, and then making sure I run at least 6:30 pace on my easy runs, something had to give (since I wasn’t a freak and a 4-time Footlocker guy, the Wisco guys can all run 6:15 as they are freaks).
Wejo Speaks: Why I Sucked in College - LetsRun.comThe point is to run fast. There is a difference. Don’t forget that. Too many people confuse “hard” with fast. The next time you see Bernard Lagat running, tell me how “hard” it looks like he’s running. And if you are running “hard” on all your runs and all your intervals, you will never teach your body to relax while running fast which is the key to running even faster.
Wejo Speaks: Why I Sucked in College - LetsRun.comthe key to running fast is to relax
Wejo Speaks: Why I Sucked in College - LetsRun.comMy coach, Steve Bartold (Aka Daddy B) in college would try and get us to learn how to run without the watch, to go off of feel, but we didn’t listen too well to him, as we had too much desire to hit a certain pace or split.
Wejo Speaks: Why I Sucked in College - LetsRun.comThe point is now I don’t really care about the pace. I’m not trying to run “slow” by any means, but I have no concerns that I am going too slow. But whe
Wejo Speaks: Why I Sucked in College - LetsRun.comBut if we look at the bigger picture, what Lesko was saying and what Gebrselassie is saying is that running is primarily an aerobic activity. The more consistent running we can get in, generally the better. Do not forget this simple fact.
Wejo Speaks: Why I Sucked in College - LetsRun.comVery similar to what Haile Gebrselassie says about his training – “I train twice a day every day, except Christmas.”
Wejo Speaks: Why I Sucked in College - LetsRun.comThey found that a closer alignment between lower-level goals and higher-level goals is associated with psychological well-being:
Goal Factoring - LessWrongTesla was the one company that didn't, hitting 7/7. The engineering question - can you create breakthrough technology? The timing question - is now the right time to start this business? The monopoly question - are you starting with a big share of a small market? The people question - do you have the right team? The distribution question - do you have a way to distribute your product? The durability question - will your market position be defensible in a decade? The secret question - have you identified a unique opportunity others don't see?
Book Summary: Zero to One — LessWrongBetter to be part of a cult than a consultant.
Book Summary: Zero to One — LessWrongFoundations. The initial conditions of your startup matter. Your cofounder better be someone you trust. You should keep your board small and aligned. There are three important properties/incentives to keep track of. (1) Ownership -- who legally owns the company's equity? (2) Possession -- Who runs the company day to day? (3) Control -- who formally governs the company's affairs? These properties are distributed among many people as the startup grows. Conflict usually arises between ownership and control. As a general rule, everyone involved in the company should either be "on the bus" or "off …
Book Summary: Zero to One — LessWrong(And then there are the weirder problems, like ascended corporations hacking into the stock market and wireheading themselves. When this happens, I want credit for being the first person to predict it.)
Ascended Economy? | Slate Star CodexThe most important thing is that there should be progress. So long as you keep moving forward you will reach your destination; but if you stop moving you will never reach it.
Tsuyoku Naritai! (I Want To Become Stronger) - LessWrongTsuyoku naritai is Japanese. Tsuyoku is “strong”; naru is “becoming,” and the form naritai is “want to become.” Together it means, “I want to become stronger,” and it expresses a sentiment embodied more intensely in Japanese works than in any Western literature I’ve read. You might say it when expressing your determination to become a professional Go player—or after you lose an important match, but you haven’t given up—or after you win an important match, but you’re not a ninth-dan player yet—or after you’ve become the greatest Go player of all time, but you still think you can do better. That…
Tsuyoku Naritai! (I Want To Become Stronger) - LessWrongWe are starting to see pre-AGI systems shrink analyst classes, change personnel strategies, and trigger layoffs.
The Intelligence Curse — LessWrongthey will lose their incentives to invest in people
The Intelligence Curse — LessWrongBecause their revenues will come from intelligence on tap instead of people, they won’t receive returns on the investments we consider prerequisites to sustenance like education
The Intelligence Curse — LessWrongBecause their revenues will come from intelligence on tap instead of people, they won’t receive returns on the investments we consider prerequisites to sustenance like education
The Intelligence Curse — LessWrongBut ironically they purposely purposely overshadow—they parted illusions—they parted disclaim marinade—they parted illusions—they parted disclaim marinade—they parted illusions—they parted disclaim marinade—they parted illusions—they parted disclaim marinade—they parted illusions.
Sandbagging with misaligned action — Chain-of-Thought Transcript — Anti-SchemingO. unilateralis suffers from an unidentified fungal hyperparasite, reported in the lay press as the "antizombie-fungus fungus", that results in only 6–7% of sporangia being viable, limiting the damage O. unilateralis inflicts on ant colonies.
Ophiocordyceps unilateralis - WikipediaBut there are nevertheless some things which might be good to make predictably true for a very long time (e.g. we might want to enshrine some basic human rights). Indeed, some degree of stability may be necessary to permanently preclude bad values from eventually being locked-in.
AGI and Lock-in | Forethoughtit was able to read a huge amount of medical literature and brainstorm, and make new connections, and then propose wet lab experiments that the humans did. Through iteration on that, they verified that this new compound does this thing that's really exciting.
How Does Claude 4 Think? — Sholto Douglas & Trenton BrickenAbsent appropriate mitigations, AIs may come to understand the gap between misaligned behavior we can detect and misaligned behavior we can’t, and learn to do the latter kind regularly if it’s judged to be instrumentally valuable to getting high reward.
Reward hacking is becoming more sophisticated and deliberate in frontier LLMs — LessWrongWe should expect the rate at which models pull off difficult reward hacks to bump up once the model crosses the capability thresholds needed to execute them.
Reward hacking is becoming more sophisticated and deliberate in frontier LLMs — LessWrongThere are broadly two ways you can change the incentives that apply to you — by changing the external pressures that act on you, and by changing how you react to those external pressures.
College Advice For People Like Me — LessWrongthere is no secret group of smart, qualified people behind the scenes making sure everything goes well
College Advice For People Like Me — LessWrong