RedwallAnalytics

@ASmithAZ @EliseStefanik I work in a high performing Bronx charter. I feel like setting a min proficiency bar for every school, charter or DOE, and vigorously changing operators of those falling short. Charters receive less funding, you could even make minimums tougher as long as enforced consistently.

Here's how New York can improve public education @EliseStefanik ✅ Expand charter schools, which research shows provide better results for low-income kids ✅Focus on the basics including phonics and content-rich curricula ✅Adopt a strong public school choice law as states such as Nevada, Wisconsin, and Delaware have done ✅Stop funding "ghost students" (i.e. hold harmless) ✅Turn off the funding spigot ✅Close empty schools that spread resources thin ✅Get out of the way of homeschoolers, microschools, and other K-12 entrepreneurs Am I missing anything?

@arpangup Our charter network in the South Bronx with 90% disadvantaged students has always taught phonics for K-1, that’s it, then we get 85-90% proficiency in 3rd Grade ELA every year, even through pandemic. There are several legs to the stool, but that seems obvious and fundamental.

The Massachusetts House just voted 155-0 to overhaul how kids learn to read, directly defying the state's largest teachers' union. Here’s the breakdown 👇 1. Despite historically leading the nation, Massachusetts is facing a literacy crisis. The latest MCAS results show that nearly 60% of students in grades 3-8 are not meeting expectations in English language arts.. 2. To fix this, the legislation pushes all districts to adopt evidence-based reading instruction, grounded in the "science of reading." This focuses on core skills like phonics (sounding out words), fluency, and comprehension. 3. The new method will replace outdated, in-effective techniques currently used in about 1/3 of MA school districts. This includes the controversial 'three-cueing' method, which encourages students to use pictures and sentence structure to guess words. 4. The bill gives the state's Dept. of Education the power to create a list of approved K-3 reading curricula. School districts will be required to choose from this list or get special approval for an alternative. 5. Despite the unanimous 155-0 House vote, the Massachusetts Teachers Association (MTA) has big concerns. They argue that a strict, top-down mandate limits their ability to use different tools to meet the needs of a diverse student body. 6. Proponents argue that the bill isn't "anti-teacher". It requires Massachusetts Dept. of Education to provide free online materials and training to help educators adapt to the new, science-backed methods. 7. The bill now heads to the MA Senate for consideration. The teachers' union plans to continue lobbying for changes that give individual districts more power to add to their own curriculum.

@JasonJournoDC I spent my life right of center, but that fell apart with rise of Trump “alternative facts”. I can’t debate untruths or 100% skepticism about things said on MSBC or by D’s, and 0% about Fox or R’s. My R family are disappointed we won’t debate, but you can’t debate falsehoods.

🚨NEW: Jon Stewart: "I’ve got people in my family that are to the right of Attila the Hun. And when people tell me, like, 'How can you platform that person on your show?' I go, 'I platform my uncle every f*cking Thanksgiving.'" "And by the way, I love him. He’s a three-dimensional human being who has qualities that I really admire — things about him." "And we’ve lost that. We’ve lost the ability to love people because we litmus test at every point, in every single moment." @DailyCaller

@gbreezy_nft @rebelEducator There were 4 parts, I seem to be paywalled out now. I read it as watching it not work in its current form for that population, and Alpha trying to experiment to fix it in real time without much success.

@lucey_david @rebelEducator This was refreshing to read, however, I left wondering what the key differences are between Brownsville and the typical model that make the others 3X+ more expensive? I'm surprised she didn't acknowledge or analyze this, though maybe she has on a different post!

Every time I post about Alpha School the selection effect counter-argument comes up So, reminder: Alpha's full of kids on financial aid No matter who the students are or how much tuition their parents can pay, the results are staggering The data's not getting skewed by "rich kids with attentive parents," the model really just works and is really that good

My pleasure to come on Dwarkesh last week, I thought the questions and conversation were really good. I re-watched the pod just now too. First of all, yes I know, and I'm sorry that I speak so fast :). It's to my detriment because sometimes my speaking thread out-executes my thinking thread, so I think I botched a few explanations due to that, and sometimes I was also nervous that I'm going too much on a tangent or too deep into something relatively spurious. Anyway, a few notes/pointers: AGI timelines. My comments on AGI timelines looks to be the most trending part of the early response. This is the "decade of agents" is a reference to this earlier tweet x.com/karpathy/statu… Basically my AI timelines are about 5-10X pessimistic w.r.t. what you'll find in your neighborhood SF AI house party or on your twitter timeline, but still quite optimistic w.r.t. a rising tide of AI deniers and skeptics. The apparent conflict is not: imo we simultaneously 1) saw a huge amount of progress in recent years with LLMs while 2) there is still a lot of work remaining (grunt work, integration work, sensors and actuators to the physical world, societal work, safety and security work (jailbreaks, poisoning, etc.)) and also research to get done before we have an entity that you'd prefer to hire over a person for an arbitrary job in the world. I think that overall, 10 years should otherwise be a very bullish timeline for AGI, it's only in contrast to present hype that it doesn't feel that way. Animals vs Ghosts. My earlier writeup on Sutton's podcast x.com/karpathy/statu… . I am suspicious that there is a single simple algorithm you can let loose on the world and it learns everything from scratch. If someone builds such a thing, I will be wrong and it will be the most incredible breakthrough in AI. In my mind, animals are not an example of this at all - they are prepackaged with a ton of intelligence by evolution and the learning they do is quite minimal overall (example: Zebra at birth). Putting our engineering hats on, we're not going to redo evolution. But with LLMs we have stumbled by an alternative approach to "prepackage" a ton of intelligence in a neural network - not by evolution, but by predicting the next token over the internet. This approach leads to a different kind of entity in the intelligence space. Distinct from animals, more like ghosts or spirits. But we can (and should) make them more animal like over time and in some ways that's what a lot of frontier work is about. On RL. I've critiqued RL a few times already, e.g. x.com/karpathy/statu… . First, you're "sucking supervision through a straw", so I think the signal/flop is very bad. RL is also very noisy because a completion might have lots of errors that might get encourages (if you happen to stumble to the right answer), and conversely brilliant insight tokens that might get discouraged (if you happen to screw up later). Process supervision and LLM judges have issues too. I think we'll see alternative learning paradigms. I am long "agentic interaction" but short "reinforcement learning" x.com/karpathy/statu…. I've seen a number of papers pop up recently that are imo barking up the right tree along the lines of what I called "system prompt learning" x.com/karpathy/statu… , but I think there is also a gap between ideas on arxiv and actual, at scale implementation at an LLM frontier lab that works in a general way. I am overall quite optimistic that we'll see good progress on this dimension of remaining work quite soon, and e.g. I'd even say ChatGPT memory and so on are primordial deployed examples of new learning paradigms. Cognitive core. My earlier post on "cognitive core": x.com/karpathy/statu… , the idea of stripping down LLMs, of making it harder for them to memorize, or actively stripping away their memory, to make them better at generalization. Otherwise they lean too hard on what they've memorized. Humans can't memorize so easily, which now looks more like a feature than a bug by contrast. Maybe the inability to memorize is a kind of regularization. Also my post from a while back on how the trend in model size is "backwards" and why "the models have to first get larger before they can get smaller" x.com/karpathy/statu… Time travel to Yann LeCun 1989. This is the post that I did a very hasty/bad job of describing on the pod: x.com/karpathy/statu… . Basically - how much could you improve Yann LeCun's results with the knowledge of 33 years of algorithmic progress? How constrained were the results by each of algorithms, data, and compute? Case study there of. nanochat. My end-to-end implementation of the ChatGPT training/inference pipeline (the bare essentials) x.com/karpathy/statu… On LLM agents. My critique of the industry is more in overshooting the tooling w.r.t. present capability. I live in what I view as an intermediate world where I want to collaborate with LLMs and where our pros/cons are matched up. The industry lives in a future where fully autonomous entities collaborate in parallel to write all the code and humans are useless. For example, I don't want an Agent that goes off for 20 minutes and comes back with 1,000 lines of code. I certainly don't feel ready to supervise a team of 10 of them. I'd like to go in chunks that I can keep in my head, where an LLM explains the code that it is writing. I'd like it to prove to me that what it did is correct, I want it to pull the API docs and show me that it used things correctly. I want it to make fewer assumptions and ask/collaborate with me when not sure about something. I want to learn along the way and become better as a programmer, not just get served mountains of code that I'm told works. I just think the tools should be more realistic w.r.t. their capability and how they fit into the industry today, and I fear that if this isn't done well we might end up with mountains of slop accumulating across software, and an increase in vulnerabilities, security breaches and etc. x.com/karpathy/statu… Job automation. How the radiologists are doing great x.com/karpathy/statu… and what jobs are more susceptible to automation and why. Physics. Children should learn physics in early education not because they go on to do physics, but because it is the subject that best boots up a brain. Physicists are the intellectual embryonic stem cell x.com/karpathy/statu… I have a longer post that has been half-written in my drafts for ~year, which I hope to finish soon. Thanks again Dwarkesh for having me over!

@ryanfazio @desegregateCT Greenwich transit districts have 30-40% parking, this seems ripe for compromise vs 10 unit apartment blocks in R6-R7 neighborhoods. The lot usage is much lower than pre covid. ctparkingreform.org

@ryanfazio @desegregateCT In my piece, suggest menu of options. Duplex/triplex/ADUs/limited resident parking permits-missing middle near 1/4ac nbhds near stations. Build over the station parking? Extended zone where appropriate to the Post Road buses. Upzone where there are bigger lots a mile away.

Wrote up why #connecticut #wlr doesn't hit spot for me, though recognize urgency for more #housing. Gathered 360k FF County parcels with State's new GIS data, and showed why "90% of parcels are 1+ acre" talking points are misleading regarding MNR stations. bit.ly/42HLLa2

I tried to show why WLR looked so difficult from a data and topology perspective from my Town of Greenwich. I hope this Legislative round can involve more compromise on both sides. I would love to see my town become Transit hub. @desegregatect @ryanfazio

And here are counts of 1 acre plus lots moving again moving progressively from same stations.

@ChadAldeman @The74 NYS lowered the passing threshold considerably for Math last year, so tempered by that, the charter district where I work in South Bronx, with almost entirely low income students, had 100% of our 101 8th graders passing the Math test in 2024, up even from 99% last year.

We can also celebrate the schools that are doing a good job. Check out my new piece @The74 on the districts that do the best job preparing kids in math: the74million.org/article/which-…

.@KevinMahnken does an excellent job summarizing the latest TIMSS results: the74million.org/article/four-i…

@StatisticUrban I must say CT middle of the pack at #20 on this list (not shown) vs #7 in median income is pretty surprising.

This is the list of shame. If you are on this list, you have some major housing reforms to work on. 8 blue states (and DC), one purple state, and one red state.

@trumbullbrod The data and software engineering community regrouped en masse at Bluesky last week via starter packs. That is the way to go.

@desegregateCT @Ct169S I meant your post screenshot at top of this chain. The previous Chair of RPA was Scott Rechler of RXR, a large developer. There is a Durst as current Vice Chair. Large past RPA donors have been Genl Contractors Assoc of NY, SL Green, Related, Suffolk Construction, Durst and RXR.

@lucey_david Hmm, please provide links to ‘this old post’! RPA has a 100 person board so it’s not ‘led’ by any individual, company or even industry But *unlike @Ct169S* we disclose our funding sources and aren’t co-founded by a realtor who personally benefits from housing scarcity!

It’s sad when electeds pander to the worst extremists in their coalitions. Especially when they know better. We get accused of being developer shills but also communists - sometimes by the same person in the same sentence! The anti-homes 169ers, bringing their best lol

@Urban_Dispatch Your FNMA chart also shows 15 years of flat or declining indices from 2005-20 under the same supply regime, and on a national level, CT has had among the lowest rates of accumulated appreciation among 50 states over recent decades, not that a revised approach doesn’t make sense.

@tvilinskis @jrojas9 Can’t help but wonder if our CT financial people can’t help here. Have votes by neighborhood near transit on issuance of tradable upzoning rights, say 6 dwelling rights per ¼ ac. ⅔ needed to approve. You have people who wanted suburban lots, but my guess is economics prevail.

1/2 @jrojas9 said "I think we have to shift the conversation to one of economics, because I think that’s what really can get people to change and think differently about this.” No need to drop the powerful moral argument, just add the economic one to its rightful place. ctmirror.org/2024/10/21/ct-…

@trumbullbrod I think they mean half mile circle around train stops as of right, regardless of already there for example. Why not offer options for good stuff you guys talk about: remove setbacks/min parking, smaller min lots, ADU’s. Maybe allow max height, flexible transit zone shapes maybe.