HNHacker News
TopNewBestAskShowJobs

kipply

701 karma · joined July 11, 2019

Programs things, circus arts. carolchen.me, twitter @kipperrii
submissionscomments
kipply··on Transformer Inference Arithmetic (2022)
oops will fix. never using names of numbers of zeroes ever again. that stuff sucks, i will express everything in scientific notation
kipply··on How the Canadian Tech Scene Encourages Finite Gameplay
oh beautiful i can just edit the title
kipply··on How the Canadian Tech Scene Encourages Finite Gameplay
There's a lot in here, but the title may be better as "How the Canadian Tech Scene Encourages Finite Gameplay"
kipply··on School or Prison
I did weekend classes in elementary school which were two hours, during which our teacher would not let us go to the bathroom. Parents eventually complained and so in exchange for being able to go pee, we were no longer permitted water during the two hours.
kipply··on Graalphp: An Efficient PHP Implementation Built on GraalVM
Describing truffle implementations as alternatives to Java implementations is not quite correct. They are always alternatives for the reference implementations (CRuby, Cpython). They're also not really about integration with Java Bytecode

Also Truffle will let you run C extensions unlike regular jvm implementations due to the polyglot nature of Graal and Sulong

kipply··on TruffleSqueak: Polyglot Programming with Squeak/Smalltalk and GraalVM
yay this is great! <3
kipply··on Laws of UX
nice compilation!

is anyone else bothered that they have to scroll below the fold after clicking "learn more" to get to the content?

kipply··on Hiring Without Whiteboards
Recurse Center has a pretty nice model where they give you a part of a program (small) to bring ahead of time (or you can pick any program you've written) and the interview is going through the code and adding a feature or two.

That kind of interview mitigates the amount of trivia and preparation-gaming while also ensuring that the applicant can solve problems as a team and is a competent programmer. It is cheatable in that one can memorize implementations of some features, though it seems somewhat difficult and not more cheatable than just memorizing a bunch of CS trivia.

kipply··on I am a model and I know that artificial intelligence will take my job
there is also rosebud.ai which creates images of clothing on (deepfake) models for businesses. It's supposed to save time and possibly money on photographers/models. Really excited for this to become viable for small businesses <3
kipply··on I Don't Want to Be a Founder
I'm afraid of Bay-Area-view-of-the-world-being-inaccurate sorta things, was kind of sad to see it in myself (though my scope was narrowed partially because my only experience was choosing between industry vs startup)
kipply··on I Don't Want to Be a Founder
This is correct (am author), bad oversight. Opinions were only applicable towards the scope you described
kipply··on I Don't Want to Be a Founder
Yah my feelings about running a startup are completely different for bootstrapped/hacked up companies (I think that's the term?)
kipply··on I Don't Want to Be a Founder
I am this close to changing it, this is very good
kipply··on Launch HN: Jika (YC S20) – Price A/B Testing for Shopify
ambition or "importance" don't always manifest in the early stages of a startup (they probably usually don't). Shopify started as a snowboard selling company and I'm sure that the founders here pitched YC a larger vision that was appreciated.
kipply··on Launch HN: Jika (YC S20) – Price A/B Testing for Shopify
Former Shopify employee, but speaking based on public knowledge;

1. Liability. A result of "raising prices by $10 is 5% more revenue" is not confirmation that raising prices by $10 is a good idea. I expect a non-trivial error rate, especially since most merchants aren't statistics savvy (forming experiments, doing risk analysis and interpreting data is hard). If Shopify causes this, it's a story. Startups will have room to develop features / legal to adjust to the liability concerns

2. Channels. Shopify is a very multi-channel platform and many stores actually rely on Facebook or Instagram integrations and physical stores rather than their online store. To be able to control prices on all these channels is an engineering challenge, and will also further dilute the value of the data. It'll also amplify concerns about pricing-inconsistencies.

3. Just a high-risk idea. Customers don't like seeing different prices. Merchants don't like running experiments that could cause them to lose money. Merchants also like being "good" to their customers by some measure, and some merchants would consider this unfair.

kipply··on Launch HN: Jika (YC S20) – Price A/B Testing for Shopify
How do you avoid customers changing their behaviours by observing pricing changes? Or when they share it with a friend and go "cute shirt for 10 dollars" and their friend is like "wat it's 15 dollars"
kipply··on Launch HN: Jika (YC S20) – Price A/B Testing for Shopify
Is there an AI(specifically deep learning) element to optimally distribute the A/B testing ratios or to suggest products/prices that may be good to A/B test? (Referencing the .ai url)
kipply··on Slate Star Codex and Silicon Valley’s War Against the Media
skipped in the article, but the deleted website specifies that the purpose of deletion is partially to ensure that if the Times were to publish an article then they'd also have to answer as to why they needlessly endangered Scott's patients and housemates.

"If there’s no blog, there’s no story. Or at least the story will have to include some discussion of NYT’s strategy of doxxing random bloggers for clicks."

kipply··on Ask HN: How Blog/side projects helped you land a job
I wrote a blog post about hugging that a google recruiter saw and emailed me about. I probably would've been contacted anyway, but the post certainly helped with making an impression!

https://carolchen.me/blog/group-hugging-theory/

kipply··on JIT Compilers Are Implemented and Fast: Julia, PyPy, LuaJIT, Graal and More
yeah my bad, I separated out the stuff about how Julia works into another post and forgot to update the title
kipply··on JIT Compilers Are Implemented and Fast: Julia, PyPy, LuaJIT, Graal and More
Haven't documented it, but I keep an oss version of my website so here's the blog part, built with Zola https://github.com/kipply/cupcake-template/tree/master/blog
kipply··on A Deep Introduction to JIT Compilers: JITs are not very Just-in-time
Thanks, I know of this stuff but I don't think I would've been qualified to form an explanation/overview of them. It would be nice to have these in a post though!
kipply··on A Deep Introduction to JIT Compilers: JITs are not very Just-in-time
I don't think I'm familiar with this (I may have come across it but didn't identify it by name when I worked with Graal?), could you provide a link? :o Thanks!
kipply··on A Deep Introduction to JIT Compilers: JITs are not very Just-in-time
A possible magic solution is to load a JIT state from a previous execution and have good manual-optimization features to be like PGO. This would give JITs best of both worlds with being able to get to a strong peak performance and also not requiring manual tuning
kipply··on A Deep Introduction to JIT Compilers: JITs are not very Just-in-time
picking colours is really hard so I decided not to
kipply··on A Deep Introduction to JIT Compilers: JITs are not very Just-in-time
https://carolchen.me/blog/jits-intro/#pogo

It's there though not very in depth c:

kipply··on A Deep Introduction to JIT Compilers: JITs are not very Just-in-time
JITs vs static compilers on a language made for static compilers is probably a contest that JITs will never win. The keys to making JITs "better" than static compilation is (by my limited knowledge) creation of languages that are made to be JIT compiled (like Java) and ensuring that tuning JITs is easier than PGO (though nothing that is currently in development seems promising to this end)
kipply··on A Deep Introduction to JIT Compilers: JITs are not very Just-in-time
Yep, class of 2019, jit to actually have a physical prom and graduation
kipply··on A Deep Introduction to JIT Compilers: JITs are not very Just-in-time
Julia compiles at runtime which makes it a JIT but it does not actually use runtime information to compile (except to know what to compile and maybe some other things) or have a concept of "hotness"

(this is in the post~)

kipply··on A Deep Introduction to JIT Compilers: JITs are not very Just-in-time
except Julia because Julia <3
Page 1 of 3Next →