HNHacker News
TopNewBestAskShowJobs

jahooma

188 karma · joined October 10, 2013

submissionscomments
jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
Yes, at least, partially. It will work, but maybe not as well as we don't parse out the function names from .svelte files.

I can add it if tree sitter adds support for Svelte. I haven't checked, maybe it already is supported?

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
We actually don't use RAG! It's not that good as you say.

We build a description of the codebase including the file tree and parsed function names and class names, and then just ask Haiku which files are relevant!

This works much better and doesn't require slowly creating an index. You can just run Codebuff in any directory and it works.

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
Yes, unfortunately, Codebuff will only read files within one directory (and sub-directories).

If you have multiple repos, you could create a directory that contains them all, and that should work pretty well!

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
Yes. Natively, the models are limited to 200k tokens which is on the order of dozens of files, which is way too small.

But Codebuff has a whole preliminary step where it searches your codebase to find relevant files to your query, and only those get added to the coding agent's context.

That's why I think it should work up to medium-large codebases. If the codebase is too large, then our file-finding step will also start to fail.

I would give it a shot on your codebase. I think it should work.

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
Cool, it's probably about the same, since we're both using the new Sonnet 3.5 for coding.

We might have a bit of an advantage because we pull more files as context so the edit can be more in the style of your existing code.

One downside to use pulling more context is we burn more tokens. That's partly why we have to charge $99 whereas cursor is $20 per month.

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
It's mainly from experience. From when I set it up I didn't have the feature to ask whether to run commands. It has been rawdogging commands this whole time and has never been a problem for me.

I think we have many other users who are similar. To be fair, sometimes after watching it install packages with npm, people are surprised and say that they would have preferred that it asked. But usually this is just the initial reaction. I'm pretty confident this is the way forward.

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
One cool thing you can do is a ask Codebuff to create these docs. In fact, we recommend it.

Codebuff natively reads any files ending in "knowledge.md", so you can add any extra info you want it to know to these files.

For example, to make sure Codebuff creates new endpoints properly, I wrote a short guide with an example on the three files you need to update, and put it in backend/api/knowledge.md. After that, Codebuff always create new endpoints correctly!

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
Yes, this is a good point. I think not asking to run commands is maybe the most controversial choice we've made so far.

The reason we don't ask for human review is simply: we've found that it works fine to not ask.

We've had a few hundred users so far and usually people are skeptical of this at first, but as they use it they find that they don't want it to ask for every command. It enables cool use cases where Codebuff and iterate by running tests, seeing the error, attempting a fix, and running them again.

If you use source control like git, I also think that it's very hard for things to go wrong. Even if it ran rm -rf from your project directory, you should be able to undo that.

But here's the other thing: it won't do that. Claude is trained to be careful about this stuff and we've further prompted it to be careful.

I think not asking to run commands is the future of coding agents, so I hope you will at least entertain this idea. It's ok if you don't want to trust it, we're not asking you to do anything you are uncomfortable with.

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
Yes, he's the lead Manifold eng. Please discount appropriately.
jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
Oh, it was a tool call I originally implemented so that Codebuff could look up the probabilities of markets to help it answer user questions.

I thought it would be fun if you asked it about the chance of the election or maybe something about AI capabilities, it could back up the answer by citing a prediction market.

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
I hear you. This is actually a foundational idea for Codebuff. I made it to work within the large-ish codebase of my previous startup, Manifold Markets.

I want the demos to be of real work, but somehow they never seem as cool unless it's a neat front end toy example.

Here is the demo video I sent in my application to YC, which shows it doing real stuff: https://www.loom.com/share/fd4bced4eff94095a09c6a19b7f7f45c?...

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
Thanks for using Codebuff! Yeah, these edit issues are annoying, but I'm confident we can reduce the error rate a lot in the coming weeks.
jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
Nope, if you go over the allotted credits on the $99 plan, then you pay per usage (with a 5% discount).

We actually ended up not charging this guy since there was a bug where we told him he got 50,000 credits instead of 10,000. Oops!

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
Haha, it's because I used to work on Manifold Markets!

Codebuff was originally called Manicode. We just renamed it this week actually.

There was meant to be a universe of "Mani" products. My other cofounder made Manifund, and there's a conference we made called Manifest!

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
We thought about making a VSCode extension/fork like everyone else, but decided that the future is coding agents that do most of the work for you.

The terminal is actually a great interface because it is so simple. It keeps the product focused to not have complex UI options. But also, we rarely thought we needed any options. It's enough to let the user say what they want in chat.

You can't point to specific lines, but Codebuff is really good at finding the right spot.

I actually still use Cursor to edit individual files because I feel it is better when you are manually coding and want to change just one thing there.

We do plan to do the SWE bench. It's mostly the new Sonnet 3.5 under the hood making the edits, so it should do about as well as Anthropic's benchmark for that, which is really high, 49%: https://www.anthropic.com/news/3-5-models-and-computer-use

Fun fact is that the new Sonnet was given two tools to do code edits and run terminal commands to reach this high score. That's pretty much what Codebuff does.

jahooma··on Launch HN: Codebuff (YC F24) – CLI tool that writes code for you
Great question!

In Codebuff you don't have to manually specify any files. It finds the right ones for you! It also pulls more files to get you a better result. I think this makes a huge difference in the ergonomics of just chatting to get results.

Codebuff also will run commands directly, so you can ask it to write unit tests and run them as it goes to make sure they are working.

jahooma··on Prediction market: We're unhappy with Firestore. What will we switch to?
>>> [Prediction markets] are not supposed to be self fulfilling prophecies. Right? And certainly the prediction market itself should not be the insiders.

We've thrown out all the rules. Why can't the prediction market be a self-fulfilling prophecy? Why can't the market outcome be decided by insiders?

What matters is what the incentives are. I say the incentives here are for you to point us to the most useful database and argue for it. That helps us.

Our incentive as insiders is to be truthful about our choice, because we want to incentivize you to help us choose a database.

jahooma··on Prediction market: We're unhappy with Firestore. What will we switch to?
Great questions!

1) There are three insiders on this decision which would be the three cofounders of this site (including me and the OP). We wouldn't insider trade because we genuinely want to know the answer to this question. The market is also tied to what we actually do (and we're not going to lie!).

2) This is both for you to predict what we will do, and to convince us. If you propose a DB and make a case for it, you can gain in expectation because maybe there really is like a 7% chance we'd pick it. So buying up shares from 0-7% is a win for you.

Alternatively, if you think that one DB choice is much better than the others, and we would be somewhat likely to figure that out, then you might gain by buying shares in that answer.

Basically, I think the incentives are good! We're using a more rigorous mechanism — prediction markets! — to do Q&A better than Stack Overflow.

jahooma··on Show HN: Create your own prediction market in two minutes
It sounds plausible, but our experience so far is that cheating is very rare. Out of hundreds of markets, we had our first clear incorrect resolution yesterday.

But the argument is not simply that people won't cheat often. The argument is that people will choose to bet in markets where the creator is trusted. Even if the majority of creators cheated, users would react appropriately and flock to the ones who don't.

It's just like how you purchase goods and services. They often have a chance to keep your money without providing what they promised. But in practice, they can't run a long term business that way.

By the way, even in our play-money currency, the market creators earn a fee on the traders' profits.

jahooma··on Show HN: Create your own prediction market in two minutes
Oracles are cool, but I humbly submit user-resolved markets as a cleaner and better mechanism.

As akrolsmir says, someone could commit to resolving according to an Oracle.

But mainly, it allows for a diversity of approaches to serve different user and subject-matter needs. Some markets can be resolved subjectively, others need to rigorously specify each case. Plus, the real world is often ambiguous. You need someone you trust to make the right judgment call.

jahooma··on Show HN: Create your own prediction market in two minutes
This is amazing!
jahooma··on Show HN: Create your own prediction market in two minutes
Ours is also market-based, rather than simply recording your prediction.
jahooma··on Show HN: Create your own prediction market in two minutes
You've actually got it backwards.

User-resolved markets is the brilliant part of our platform, but most people are not able to figure that out.

jahooma··on Show HN: Create your own prediction market in two minutes
Prediction markets : The Delphi method :: Classical liberalism : Liberalism
jahooma··on Show HN: Create your own prediction market in two minutes
Today's the launch day for Manifold Markets, you can read our announcement here: https://manifoldmarkets.substack.com/p/above-the-fold-manifo...

We've been working for the past two months to make the easiest-to-use prediction markets! We want to make creating and trading in prediction markets as frictionless as using Twitter polls.

It's been a very exciting and fulfilling project to work on given the dedicated fans of prediction markets that have already joined our community.

Happy to hear any feedback you have on our site!

jahooma··on Mantic Markets – user-created prediction markets
Hi guys -- Mantic Markets co-founder here! Thanks for checking out our site!

Most startup ideas are counter intuitive, but I appreciate the skeptical response here.

The case for user created and resolved prediction markets is that it's simple, clear, and actually what traders want.

For example, Zvi, who writes on Omicron and consults on prediction markets, recently blogged that the ideal way to resolve markets is to have Scott Alexander be the judge. https://thezvi.wordpress.com/2021/12/02/covid-prediction-mar...

Reality gets messy and judging outcomes ultimately comes down to a person's priors and expertise. We envision a world where experts that you trust run markets on topics that you trust them on.

Right now, we have launched a play-money version of our site, and are exploring options such as moving into crypto. I will say that we love how easy and fun it is to participate with play-money, and believe that the results can be accurate.

Imagine if your favorite Twitter and Substack personalities mixed in prediction markets with their regular tweets and posts to help their analysis or make a point. It's like a Twitter poll, but with a system for accountability that drives accuracy!

We think a super-accessible site that makes accurate forecasting fun can lead to a profound impact on our knowledge and guide us to a better world. Thanks for reading!

--James

jahooma··on Ask HN: Has Y Combinator sent emails already?
Thanks for posting, I thought we had already been rejected by not receiving an email...

I'll keep waiting then.

jahooma··on Salt Lake Screaming Eagles, a football team controlled by fans with smartphones
It certainly fits the "Screaming" Eagles theme:

Salt Lake Screaming Eagles, a football team controlled by fans with megaphones

jahooma··on Amazon Go
Actually income inequality was highest in agricultural times (higher than today), and hit its bottom in 60s or 70s for the US/UK, well after industrialization. Life expectancy and health inequality is near all-time lows right now. Information/education inequality is also at all-time lows, with more people in college than ever, and near universal access to the internet. Entertainment inequality is at all time lows.

Just wanted to note that we're really not doing too badly with advancing technology. And if you measure actual poverty as compared to relative poverty, which makes more sense to me, we are doing even better. Have a nice day.

jahooma··on A maths question for Singapore schoolkids
August 17, assuming Bernard would have said something different if he knew Albert's first comment was impossible.
← PreviousPage 2 of 3Next →