Hacker News Reader

KeatonDunsfordon Context is the bottleneck for coding agents now

This is precisely how I go about my usage pattern with Cursor that I already have, I structure my repo declaratively with a Clojure and Nix build pipeline so when my context maxes out for a chat session, the repo is self-evident self-documented enough that a new chat session automatically has a heightened context

- - kae3g

KeatonDunsfordon Context is the bottleneck for coding agents now

Here's a project I've been working on the past 2 weeks and only yesterday did I unify everything entirely while in Cursor Claude-4-Sonnet-1M MAX mode and I am pretty astounded with the results, Cursor usage dashboard tells me many of my prompts are 700k-1m context for around $0.60-$0.90 USD each, it adds up fast but wow it's extraordinary

https://github.com/foolsgoldtoshi-star/foolsgoldtoshi-star-p...

_ _ kae3g

KeatonDunsfordon At high pressures, potassium adopts a mix of crystalline and liquid structure

I recently enjoyed reading Dr. Gerald Pollack's "The Fourth Phase of Water: Beyond Solid, Liquid, and Vapor". I wonder...

https://www.amazon.com/dp/0962689548/ref=cm_sw_r_cp_awdb_t1_...

KeatonDunsfordon Tim Berners-Lee wins Turing Award

Help build Urbit.

KeatonDunsfordon Own your identity (2011)

"In an Urbit world, your data is no longer trapped in a jumble of proprietary servers. Your urbit is a permanent, versioned, typed archive the size of your digital life."

https://urbit.org/posts/overview/

KeatonDunsfordon Urbit in 2017

Not well, yet. But that's the plan. All in the blog post.

KeatonDunsfordon Urbit in 2017

This is probably our best general overview: https://urbit.org/posts/overview/

We just did a podcast with Software Engineering Daily also: https://softwareengineeringdaily.com/2017/01/20/urbit-with-c...

But the blog really highlights well what we're trying to accomplish in the near-term.

KeatonDunsfordon Astronaut – YouTube videos with almost zero previous views

Just randomly stumbled upon this and almost teared up a bit. Absolutely beautiful, and brilliant.

KeatonDunsfordon Show HN: React Native Express – Learn React Native Through Interactive Examples

Omg perfect timing. I was just looking for something like this. Y'all are the best.

KeatonDunsfordon YC W17 Invites/Rejections

Rejected. Get it though, as my thing is both really (probably too) big and really early. The grind continues though.

KeatonDunsfordon Guidelines for Edtech Products

The model I've come up with leapfrogs government entirely and goes straight to users.

KeatonDunsfordon Guidelines for Edtech Products

With all due respect, it's a total fallacy that edtech has to be different from other industries. There's no physical force in the universe forcing education to be any certain way. So if one way is provably better, we should just build it, and deal with the obstacles when they come up later. Kids, both privileged and not, have been dropping out for years because they're tired of things. I'm building this for them.

To quote Balaji Srinivasan of 21 & A16z:

"Don't argue about regulation. Build Uber. Don't argue about monetary policy. Build Bitcoin. Don't argue about it. Build the alternative."

KeatonDunsfordon Guidelines for Edtech Products

Sounds good. Shoot me an email at keaton@bloomv1.org. Other social media info is in my HN profile page.

KeatonDunsfordon Guidelines for Edtech Products

If you look over my application linked in the OC, you'll see our number one design goal is to connect users physically. There's something incredibly powerful about the social experience that school provides like you're describing that no big edtech platform today has made an explicit design goal. In short, we'll have to "design for trust" a lot like Airbnb did (see Joe Gebbia's TED talk). The locations could be anywhere from a local Starbucks to a public library to the nearest college campus. If educational content could be optimized for the smartphone, a lot like Snapchat has optimized the Snapchat story as their primary unit of content on mobile, then any place could become a sort of classroom. So long as you're surrounded by kids like you learning and making things with you that you care about. Meetup.com has accomplished this to some extent for working professionals, but we don't think it's specific enough towards specific projects in the context of youth education.

This is going to be a really hard problem to solve overall from a design perspective. And like you hinted, it will be even harder for us to do this for kids in poor or rural areas in contrast to rich kids in urban areas. But we're going to figure it out somehow. If the timing is right, self-driving cars and improved cheap public transportation will help this tremendously. Our app could then in theory automate the entire process of getting a group of kids connected somewhere safe and public. The value and safety of those kinds of experiences could be built upon even further if you add in a kind of crowdsourced mentorship experience on top of that, like Uber for chaperones in a way. If you're a local software engineer or cinematographer or biophysics researcher, you could give back to your local community by helping out local kids who want to contribute in your field. Now all the sudden everybody can become a teacher, which feels really good. Like Quora answers, but in person.

KeatonDunsfordon Guidelines for Edtech Products

I've been incredibly inspired by YC and only mean for my criticism to be constructive. But by naming their edtech initiative "Imagine K12", they've already lost.

I care as much about solving education as they do. I think the basis of my argument is factually correct. Just look at the data.

Why not treat education like any other disruptable industry?

KeatonDunsfordon Guidelines for Edtech Products

Nope, which is why I applied.

And I agree in the sense that I believe in YC more than anybody. But I disagree in the sense that no edtech solution to date is dramatically changing the world as much as a Facebook or Uber, so something must not be working. And I'm calling out Coursera, Udacity, Udemy, EdX, Khan Academy, Codecademy, Vidcode, Jumpcut, MakeSchool, AltSchool, HackClub, codings bootcamps and regular school itself. Honestly, the best solutions right now are in fact YouTube, Reddit and Hacker News.

KeatonDunsfordon Guidelines for Edtech Products

But that's my point. Uber and Airbnb and Bitcoin didn't confine themselves to laws in the beginning. They just built something that worked. Yes, now that they are big they are hitting legal obstacles (like NYC's ruling on Airbnb rentals today, a top HN post). But they are also now known, and so users are aware of the potential innovation that the current state of law is inhibiting. And that increased demand for these kinds of inevitable new products and services will lead to faster change over time.

KeatonDunsfordon Guidelines for Edtech Products

You may be right. But if I'm right, I'll help more kids faster. And that's what I care about. And I think I am right, because I see what I'm talking about every day.

With the exception of a few incredible teachers and mentors out there (I myself had several), education today isn't genuinely exciting for most students. Many teachers are truly great people and are valuable to students. But are they a necessity, or a luxury? Either way, in order for educational content to stay current in today's system, teachers need to be taught new information first before they can then teach their students. Physical teacher-to-student teaching is an O(n^k+1) solution that will never be in every city let alone the developing world. If -- and that's a big "if" -- learning could come from a single software-based source of information in a way that was just as good or better than physical teaching today, a world-class education and credit for it could be accessible everywhere. All a student would need is a smartphone, which is relatively nothing compared to the current cost of education. Look at Duolingo as a great example of this for language learning and language-fluency credentialing. And on top of that, is the role of the teacher to in fact teach, or to guide and mentor? Great mentorship is invaluble, but that can also come from anybody great even if they're not in a formal teaching role.

Most learning today, whether technology, science or art, happens through books and YouTube videos and students teaching each other anyways. This peer-to-peer model works. And because through this model, young, radiant students are the primary source of content delivery, this model is also addictively social and exciting. All you have to do is remove the system and let kids do their thing. There's a reason products like Reddit are so popular -- it's an Internet ecosystem for youth all over the world.

Almost everyone who's been largely successful in any field has been self-taught. This isn't a coincidence. Those who truly love what they do enough to commit themselves to becoming world class by definition transcend the limits imposed by conventional educational standards. The problem, then, is not a teaching problem, but an exposure problem and an excitement problem to help kids find their passions and solve problems with friends in person who they love. Could an edtech platform be as viral as a Snapchat, Twitter, Netflix, Twitch Reddit, Buzzfeed? That's what I'm set to find out.

Edit: For context:

Harkness table https://en.wikipedia.org/wiki/Harkness_table

Sugata Mitra: Kids can teach themselves https://www.ted.com/talks/sugata_mitra_shows_how_kids_teach_...

The best prep school in the world Phillips Exeter Academy, where I went, ironically is designed around kids teaching themselves. And parents pay $40,000+ per year for it.

KeatonDunsfordon Guidelines for Edtech Products

To be honest, I'm extremely disappointed in these guidelines, given that this is coming from YC. Why? They divert focus away from doing whatever it takes to make something incredible for users: underserved students in a failing system that's becoming exponentially costly and disconnected from our modern society. Even if that means challenging convention and breaking some rules. Think of where Airbnb, Uber and Bitcoin would be if they had stuck to standards, to name a few big examples. This post goes completely against the "naughtiness" YC (PG) has always looked for in founders. For some reason edtech to date has been an exception to this rule.

Disclaimer: I just submitted a YC W2017 nonprofit application as the solo-founder of Bloom, a new education system powered by software. I'm not sure if this comment will help or hurt me if I get responses, but I think something needs to be said. I put my app up on Dropbox for you to check out if interested. Still searching for a great cofounder. If this is an issue you really care about, feel free to DM me on Twitter. I'll fly out anywhere. Bloom YC W2017 application: https://www.dropbox.com/s/1szvs6o4qs15djj/KD%20%26%20Bloom%2...

The cost efficiencies of conventional public and private education today are abysmal. Teachers don't scale. Campuses don't either and are physically isolating. And Common Core, courses, grades, tests, and degrees are arbitrary practices that could by replaced by something completely different, something designed for the digital age. Something not only better than public schools, the most prestigious private prep schools and the Harvards and Stanfords, but something that could scale to everyone in the best and worst of areas of the world. Education is a lifelong right that everyone clearly needs and deserves as technology continues to advance exponentially.

Yet for some reason, edtech hasn't been able to look beyond the current state of things and build something completely new. There isn't an "Uber" or "Bitcoin" of edtech. And I believe the reason why is solely because of posts like this: advice that places too much importance on working with inefficient systems, industry best practices, laws and trying to fix a broken system instead of ignoring all of that unimportant stuff to just focus on users and build a new one from scratch through software. As a result, current edtech solutions are fragmented, inflexible, slow, general and boring, just like physical schools.

So what will education look like in the future? In my opinion, it will be a cohesive system that’s modular, peer-to-peer and mobile. That’s the only way an edtech platform can both scale to ->7.5 billion users and support arbitrary specializations adaptable to the future.

I think a good parallel to the approach I'm talking about is what Urbit has been trying to do to re-decentralize the Internet. Instead of trying to solve each problem one at a time, they've built a cohesive system from scratch on top of the old, broken Internet, which solves all of its problems in one go. This is the approach I think edtech needs to take, and ignore everything else.

Yes, clear product vision and implementation ease are important for edtech startups. But you could say the same about Uber and Airbnb. Yes, viral current content is important. But you could say the same about Netflix and Twitch. Yes, flexible, optimized pricing is important. But that can be solved in other creative ways. Yes, exceptional customer support, respect for privacy and organization transparency are vital, but that's expected of any great YC startup.

I'm just shocked software hasn't eaten the education system yet, because ultimately the quality of kids' entire lives are on the line. I would know.

KeatonDunsfordon Silkworms that eat carbon nanotubes and graphene spin tougher silk

Silkworm son: "What's for breakfast?"

Silkworm dad: "NANOTUBES."

KeatonDunsfordon I Am Sam Altman, President of YC Group. AMA

Hey Sam. I'm considering applying with my nonprofit edtech startup (a new crowdsourced education system that's peer-to-peer and mobile -- built around software, like Watsi). I'm a solo-founder still currently building the MVP.

Quick question 1: Is there a "Delaware C Corp" equivalent for the incorporation process/legal structures of nonprofits by state, or does it not matter?

Real question: Is there something to be said about founders waiting until they have users and initial growth to then use the YC opportunity to transition towards growing their company? In other words, if a startup is only getting one shot to go through YC, generally is finding initial product-market fit harder, or is growing the company?

Granted, the latter can't happen without the former happening first. But it would seem that for some founders that have genuine insights into real big problems, it would suck to waste the opportuniy of YC just "checking the boxes" of building their MVP, which they could know how to do already from PG's essays/YC's blog/Startup Class -- whereas they could be getting genuine advice on problems unique to their specific domain problem if they just waited and applied later.

KeatonDunsfordon Why Ruby is an acceptable Lisp (2005)

For a 2005 article, this is totally awesome, as well as that Steve Yegge response. But in 2016, it's hard for me not to get excited about what's seeming like Clojure's upward trajectory into the mainstream and want to jump on board 100%. To paraphrase Rich Hickey, Lisp and immutable functional programming totally rock, and the JVM and Javascript just reach (ClojureScript). As our systems get bigger and more quality is demanded from our software, it'd be so cool if we could get behind this as a standard, in all the growing technology fields -- web and mobile (Om.next for React and React Native), big data, AI/machine learning, Dockerization/containerization and even decentralized stuff (Pelle Braendgaard's Cloth library for Ethereum is a start, and I was just talking with some awesome people in the Clojure Slack channel working on Boot tasks for IPFS, and soon the Golem Network project will be launched which will be super dope). Hell, even designers can start designing in Clojure now. It's just Simple, and as the community grows, rallying behind it is only going to get Easier. People were raving with React Native about how JavaScript developers could now all the sudden start coding mobile apps. How cool would it be if there was just one awesome language, the one true immutable functional Lisp, where everybody in tech could understand each other and collaborate? I genuinely think this is Lisp's revenge (shoutout David Nolen!).

It's been tough for me as a new developer to learn this stuff without nearly as many resources as there are in the Python and Ruby communities for beginners, but there's no question this is the (near-term) future. Plus, a lot of that stuff just complects everything anyways and doesn't have to apply anymore (read: mutable state, no Value on Values).

Let's keep it Lispy. ;)

KeatonDunsfordon Ask HN: What should I ask Elon Musk for HTBTF?

Relevant for both YC & Elon: Since it seems like every company trying to hyperscale is experiencing a desperate shortage of tech talent, how can we get more unexposed people excited about technology? If Elon didn't have an engineer parent, what could have been another avenue for him to have been exposed to technology, and how might we be able to target those entry points globally to train more great founders and engineers?

KeatonDunsfordon How I Built a Custom Camper Van (2015)

This is amazing. Instead of getting an apartment with my cofounder and office space for our engineers, I'm just going to have us get a fleet of these things. Would be so dope. Be anywhere -- SF, South Bay, Berkeley, LA, New York. Live the dream now.

KeatonDunsfordon An act to amend Financial and Corporations Codes relating to digital currency

Can imagine this will be impossible to force with the rise of decentralized exchanges for BTC and ETH like Bitsquare and EtherEx. But still...

"The bill would prohibit a person from engaging in the digital currency business without enrolling in the program and would prohibit the conduct of digital currency business through an unenrolled agent. The bill would require a person seeking enrollment to pay a nonrefundable fee of up to $5,000... The bill would also require the person to provide fingerprints and would authorize the commissioner to deliver the fingerprints to law enforcement agencies."

Fuck this.

KeatonDunsfordon Coding is not a Magical Cure-All

Anyone working on any cool edtech-related ideas/side projects to create something cooler than the stuff the author is referring to?

KeatonDunsfordon Ask HN: What has changed since reading HackerNews?

No doubt. HN naturally selects for smart people and relevant stuff. And no ads!