HNHacker News
TopNewBestAskShowJobs

yaseer

1,886 karma · joined January 2, 2015

Co-founder axiom.ai, YC W21 https://axiom.ai/

Outside of work, I'm extremely interested in automated theorem proving, the lambda calculus and its applications to computational biology.

An out-of-date blog explaining my old research is here: https://yaseersheriff.github.io/

submissionscomments
yaseer··on Ask HN: Advice after a decade trying to get into YC
(Applied twice, got in second time)

Don't hide anything.

Applying again is not a problem, from what I have seen. One thing you'll see in Paul Graham's writing is that persistence/resilience is the most important attribute of a founder. PG's thinking has defined YC's culture, and that does not seem to have changed.

Secondly, don't do things to get into YC. Do things because they are useful for your business.

Assembling a strong founding team is the foundation of your business - you should do this because it's intrinsically useful.

Don't focus on getting into YC - focus on building a product people want, and a company to seize that opportunity.

Ironically, if you stop devoting mental energy to getting into YC, and devote that into your startup, this will make getting into YC easier.

yaseer··on Millet, a Language Server for SML
I also didn't know how to apply the concepts at the time, it seemed more academic than practical, but I had a different level of enjoyment.

Having been a script kiddie teenager, just hacking together stuff that worked, functional programming seemed like some beautiful alien artifact.

I didn't really understand it, but it seemed elegant and enigmatic, which made me want to learn more.

yaseer··on Millet, a Language Server for SML
Cool project!

I always thought SML was a great language for teaching computer science and functional programming.

A lot more fun than learning Java. With ML, you got glimpses of aesthetic beauty in code.

yaseer··on Learnings of a CEO: Wade Foster, Zapier
There's a popular misconception on HN that YC is all about VC funding.

I used to think this too, before doing YC and realising YC emphasise funding as a tool, not as a goal. You raise as much as you need to achieve your goals.

Zapier really embody that lesson writ large - they didn't take additional funding because they didn't need to. I hope their success emphasises that lesson.

yaseer··on Google geofence warrant could help sentence murder suspects to death
https://web.archive.org/web/20220728183340/https://www.forbe...
yaseer··on I Am a Spammer. AMA (2009)
It's possible this post may be a significant embellishment of the truth, rather than being entirely true or made up.

He could well be a spammer, perhaps a small time one, that romanticises being the big-fish criminal living an implausible life on a boat in the Caribbean.

Perhaps the life of a spammer is more mundane, and those parts of his AMA are true, but he felt the need to jazz it up a bit.

yaseer··on The endless search for a crypto use case
For the record, I don't think crypto is useless, I just don't think it's hitting that bar for a transformative technology. It's proponents certainly do describe it as potentially transformative or revolutionary.

Yes, I'm aware the dates of invention don't quite align, and the time window is cherry-picked for dramatic effect. But it's food for thought.

Nobody was debating the utility of cars and phones 14 years later. With crypto, it's like we're still looking for something as fundamental as an essential use case.

yaseer··on The endless search for a crypto use case
I'm reminded of the pictures of New York in 1900 and 1913. Within those 13 years, the car had changed the world.

https://www.businessinsider.com/5th-ave-1900-vs-1913-2011-3?...

The iPhone and smartphone revolution clearly did the same, last decade.

We're now 14 years into Bitcoin. Is it ever likely to change the world the same way?

yaseer··on Why is it so hard to give Google money?
My experience correlates - Google ads for early stage startups and SMEs is a terrible product.

Clearly, it has been optimised for businesses with significant ad budget.

But I do wonder - are they not leaving significant SME money on the table?

You would think that's a large enough SME market globally to support building an SME product experience that isn't so terrible.

yaseer··on Manhattan rents cross $5k threshold for first time
This would have been true when GBP was strong, but it's now trading at 1.18 USD. The glory days of 2.05 USD from 2007 are long gone, sadly!

Even in Kensington and Chelsea, average rent of £2.5K or even £3K GBP is only $3K-$3.5K USD these days.

https://www.homeviews.com/renting/average-rent-in-london-for...

yaseer··on Your compliance obligations under the UK’s Online Safety Bill
> instantly erase someone from your perception in person to person interaction.

As easily as you can block a message, so too can abuser create a new account.

Online harassment can be as persistent as in-person harassment - there's countless stories of people being continuously harassed. If it were so easy to block people online 'cyberstalking' wouldn't be such a big problem in the same area.

yaseer··on Your compliance obligations under the UK’s Online Safety Bill
> An adult getting abused online is like getting third degree burns because you laid your hand on a hot burner and refused to take it off

This analogy is broken on quite a few levels. When did the internet become a hot burner?

The internet is not designed to get hot - it's not a cooking device. If your laptop consistently got so hot it burned your lap, you'd try to fix it, not say "of course it does that".

> If people saying mean things to you is disturbing your groove so much you're calling it "abuse" maybe you should stop reading them

There are laws governing abusive behaviour in person. Our society does not just advice people to "just not listen".

Why would you expect different principles to apply to online and in-person behaviour?

https://www.cps.gov.uk/crime-info/verbal-abuse-and-harassmen...

yaseer··on Show HN: [NSFW] Diffusion models for porn generation
If it exists, there's porn of it. And a relevant xkcd.
yaseer··on Copilot sells code other people wrote
Technically, programmers search, copy and modify code all the time.

One might argue copilot puts into software an algorithm that humans are already doing. Software like that is usually inevitable.

Still, it sucks there's no benefit for the contributors.

The most ethical thing I can think of is some kinda 'Spotify-like' revenue sharing model, based on how often their code is used by others. Not that they'd ever implement that if they can get away with it!

yaseer··on Ask HN: I created a news shortening algorithm and am not sure how to utilize it
Comparisons to competition are just a way to map out the market, and where you may fit.
yaseer··on Ask HN: I created a news shortening algorithm and am not sure how to utilize it
If it's hard to tell the difference, it'll be hard to sell the difference. (Cheesy rhyme intentional)

But seriously, you could probably do some research to work out your solution's strengths, relative to existing solutions.

When you know what it's strengths are, try to find people who want those strengths.

Ultimately, you're asking the question 'is there a market for what I've built?' - but you've phrased that question differently.

yaseer··on Ask HN: I created a news shortening algorithm and am not sure how to utilize it
Can you articulate why this is different to other text summarising solutions?

It may be that yours is easier to integrate than using AWS APIs[1], or performs better than what's available on say, npm[2]. It may be that your algorithm is designed specifically for news articles.

If you can articulate where this fits into the market of other solutions - that will help inform how best to utilize it.

[1] https://aws.amazon.com/blogs/machine-learning/part-1-set-up-...

[2] https://www.npmjs.com/package/text-summary

yaseer··on What we learned in studying the most effective founders
This reads more like a generic SEO page, stating platitudes, rather than a novel analysis by one of the largest, most influential companies in the world.

If Google considers this quality content, is it really surprising their search results increasingly return non substantive answers to questions?

yaseer··on Internet Explorer 11 has retired and is officially out of support
Microsoft's leadership and strategy has evolved under Nadella, at least enough to produce quality products like VSCode, and 'good enough' products like WSL.

I say this as someone who has historically been very anti-Microsoft, that used to do all my work in Linux. Modern Windows with VSCode, running WSL, is something the Microsoft of the IE6 era would never have produced.

yaseer··on Mental model examples and their explanations
You have to sift out a lot of junk from this article, to get at the useful points. It would benefit from some editing.

> “It’s usually easier to make something 10x better than it is to make it 10% better"

This is not 'usually' the case. Usually, the exact opposite is true.

yaseer··on Show HN: Vantage Autopilot – Save on AWS EC2 Costs
> This seems like something you could build over a weekend

I think you've succinctly illustrated why software estimation is notoriously unreliable.

Those estimating do not yet understand the problem.

yaseer··on Show HN: Vantage Autopilot – Save on AWS EC2 Costs
This is really cool!

If you do what you say you do, this is a lot of value to be had relative to the costs.

We're coming to the end of our cloud credits and I was thinking about next steps - definitely giving this a shot.

yaseer··on Owl: OCaml Scientific Computing
I wasn't aware that ahrefs was supporting Ocaml projects.

Jane Street obviously comes first to mind. Good to see more companies develop the ecosystem. It used to be a very niche, esoteric language.

I discovered Ocaml whilst learning Isabelle and suddenly fusing OOP and functional programming seemed elegant, whereas Scala had made it seem scary.

yaseer··on Ask HN: What are some good inspirational and uplifting books?
Seconded.

It shall bring you down emotionally, then build you up again - but stronger.

yaseer··on Ask HN: Important nonobvious startup/business lessons you've learned?
> If the idea / product is any good, it'll probably catch on quickly and sales will be easy.

I would respectfully disagree with this.

There are a few startups (e.g. Facebook), whose market timing is so perfect, that it's a sure-fire hit from the start.

But I have known so many companies that had founders struggling to make sales, as their timing was off. Then some market shift occurs, and what once was hard becomes easy.

COVID is a great example of something which induces a seismic shift everywhere. Remote work companies for example, would have suddenly found sales an order of magnitude easier.

yaseer··on 100-year-old Brazilian breaks record after 84 years at same company
Funnily enough, my own dad had a quirky idiolect where he referred to every family member with 'the' definite article: 'The wife', 'The son', 'The daughter' etc. Your mistyping sent me down memory lane!
yaseer··on An automatic theorem proving project
Exactly - machine-formulated ATP often ends up like machine-code, very similar to the proofs of Russell's Principia Mathematica. Extremely low-level formalisms great for procedural reasoning, but poor for human understanding.

I would speculate Gowers is looking at higher-level abstractions that capture the essential semantics mathematicians are interested in - very much like a higher level programming language that humans understand.

yaseer··on An automatic theorem proving project
> They have a hunch, pursue it

There's a lot to unpack in just this.

Can we define an algorithm to generate hunches? (mathematical intuition).

Given infinite hunches, can we define an algorithm to order hunches by priority, to search?

That's the kind of question they'll be thinking about using the symbolic logic/classical algorithms in GOFAI approaches.

yaseer··on An automatic theorem proving project
For a mathematician, it's not the output of pruning the search space, but understanding how the search space is pruned.

Understanding how ML models work is a notorious problem.

A classical algorithm can be formally understood better - I think it's this mathematical understanding Gowers is looking for.

yaseer··on An automatic theorem proving project
I don't think Gowers is trying to build a better theorem prover.

An AI researcher would be trying to build a better prover.

Gowers is a mathematician looking for mathematical insight.

This project's success would be based on how well it answers those fundamental questions he posed (i.e How does a human generate proofs, given no algorithm exists to solve the halting problem?)

The output might be more like meta mathematics, with insights provided by software, rather than ML software optimised for a goal.

← PreviousPage 3 of 14Next →