HNHacker News
TopNewBestAskShowJobs

OmarShehata

1,448 karma · joined March 23, 2012

Graphics programmer. I like telling stories.

https://omarshehata.me/

submissionscomments
OmarShehata··on Show HN: Building a GPS receiver
Amazing! I also had exactly the same experience that led me to research this a few years ago, realizing that:

- GPS works even in airplane mode (while on a literal airplane) - It works without cell service, or wifi, or anything - The United States of America controls the GPS constellation, and they can (and have!) turned off GPS off certain regions at will when necessary (which has prompted other countries to launch their own GNSS constellations) - GPS satellites don't send down a location, they only send down time

I think it's a really fun exercise to do this with data you receive on your phone. Your phone has a direct link to satellite.

(side note: I recently learned the basic principles of star navigation, and while it is a completely different mechanism, it also relies very much on keeping accurate time, which I thought was a fun symmetry!)

OmarShehata··on AI-generated sad girl with piano performs the text of the MIT License
At that point, is it AI generated?? That feels like an entirely different category to me

(like it's sort of no difference than paying someone to voice something and share it)

I think the stuff that is completely generated with no human in the loop is a different category for me because it can be used for things at scale like, bots on social media, or ads in a podcast generated just for you, etc. As long as there is still a human in the loop making the editing decisions, it feels not categorically different from the world we have today.

OmarShehata··on Unending bubble wrap: a pointless activity to spend a few satisfying minutes
I was totally expecting it to support infinitely recursive zoom! (as seen in https://news.ycombinator.com/item?id=39799755)
OmarShehata··on Game of Life, simulating itself, infinitely zoomable
I was a big believer in Haxe, it seemed perfect. The main issue I and many people had at its inception was that the ecosystem was very fragmented. The promise was "write once, run everywhere" (which is what we had with Flash), but each platform you wanted to actually compile on was a whole process, often undocumented, with lots of caveats, and having to hunt for third party community wrappers to do the native integrations you want etc.

This was ~10 years ago, I'm sure the current state of the ecosystem is very different (no idea), but this is the reason why I was never able to adopt it, and I heard the same thing from dozens of Flash creators at the time.

OmarShehata··on How to embrace asynchronous communication for remote work
> I just can't figure out what the fixation is

I see this happen over and over: there's a new system change, some things don't work as well, so people revert & try to cling to the old system.

This is a bit of a fallacy that comes from looking at only what is lost, and not the total tradeoff of what is gained & lost. I think it's not unreasonable to fixate on sync meetings, because your procedures & culture have been built around it. But I think it'd be more productive to "roll forward" and figure out how to make the best use of the new system, and plug in any gaps it has.

OmarShehata··on Quiet-STaR: Language Models Can Teach Themselves to Think Before Speaking
This has always been true of anything anyone has ever googled or looked up on stackoverflow

I copy paste code from stackoverflow all the time. I used to agonize over making sure I fully understand every line it's copying. Now I have the discretion of making that decision: sometimes it does really matter, sometimes all you need to know is that it produces the right result for your limited use & test case of it. (it's no different than relying on a 3rd party library in that way)

I think we need to apply the same discretion to LLM output. The answer "it depends". Sometimes using its output blindly leads to disaster. Sometimes using it without fully understanding all the details is a great way to make progress.

OmarShehata··on Quiet-STaR: Language Models Can Teach Themselves to Think Before Speaking
I want to second this:

> Use if for things where you think "this is probably easy but I have no idea how to do it"

I had exactly the same reaction as OP (LLM's suck what's with the all the hype). These people are using it differently. For me it's often something like, asking it to put together a specific sequence of matrix transformations in ThreeJS or some other library.

This is not a difficult task but it's often one I waste a lot of time getting right. It's sort of about finding the right level of abstraction you need to ask it.

OmarShehata··on Eloquent JavaScript 4th edition (2024)
I very much concur that it is better NOT to introduce TS when explaining JS fundamentals. I've seen smart engineers with a C++ background get tripped up on and very confused working with TS, because it's not clear to them what concepts are "language fundamentals" and what concepts are "the TS transpiler".

(Like expecting that just because you declared something as type X, that it guarantees at runtime it will always be type X, but it won't. You may get data from an API that _says_ it returns type X but the contents don't match. That can be valid code that compiles and has weird runtime behavior)

OmarShehata··on Valid U.S. visa didn't stop these Chinese graduate students from being deported
The most important part is that this is the first time we're seeing the effects of a new law, (1) giving CBP officers discretion about who to deny, even if they have valid paperwork, (2) not requiring them to give a reason why they were denied

> A critical factor in the recent deportations may have been a directive issued in May 2020 by then-President Donald Trump. It gives CBP the authority to deny entry to Chinese graduate students and postdoctoral scholars who have been supported by Chinese institutions carrying out a 2017 government edict that endorses acquiring foreign civilian technologies that could also have military uses. However, there’s no official list of such entities or of the relevant technologies.

OmarShehata··on Show HN: Replay your typing in a few lines of JavaScript
This is great!

I was just talking to a coworker today about "libraries" vs "templates". You could have made this a library, but I think it's actually way nicer for it to be code I can copy/paste, so I can integrate it however I want. Some things are better as templates that I can copy.

If I have a weird use case or want to integrate it deeply into something else, the library doesn't need to accommodate that.

(the tradeoff of course is, if you update the library, my copied code won't update, so obviously this doesn't work for everything, but I think it works for some things!)

OmarShehata··on Falsehoods junior developers believe about becoming senior
The good news for everyone is, you can't just make up a good reason out of thin air! (unless you're lying). There's either a critical business need, or there isn't.

This of course requires this senior engineer to be business-savvy enough to discern when there's a legitimate business reason given, and at big companies that may require knowledge of politics among the higher ups which may be opaque to most folks.

OmarShehata··on Scientific journal publishers and editors say they are being offered bribes
Interestingly, one of the comments on the article makes this exact point ("Is this actually harmful to the scientific community?" "No one in the real world of scientific progress is getting fooled by a Mad Libs paper in a Mad Libs conference proceeding.")

The real harm here may be moreso outside of the scientific community. Consider why they are paying to get these articles published in the first place. I thought this was well articulated in another comment there:

"These papers are the tools by which fraudsters advance up the acaddemic ladder. As a result, billions of public money get diverted to fund this fakers rather than actual science and scientists.

On top of that, all these fake papers poison the waters and confuse decision makers and the public all over the world, resulting in bad policies that cause more harm."

OmarShehata··on NASA loses contact with Ingenuity Mars helicopter
"killing each other over whose god is correct" is a profound misunderstanding of humanity. That's not why humans kill each other now, or in any time in history. It is and always has been simple & rational game theory: fear of the other tribe killing you for your resources, or just simply because they're afraid you might kill them for _their_ resources.

For example, this is (one of) the core reasons of why peace in Israel/Palestine is so hard (each side fears accepting sovereignty of the other side means they will continue to encroach and wipe them out).

Humans are pretty clever and ingenious at all levels

OmarShehata··on The Serendipity Machine (Notes on Using Twitter)
I've been using twitter for 10 years, and it's generally been "OK" (but still net positive, it helped me get an internship and a job). Last year twitter became an incredibly positive thing in my life (in spite of the exodus and things generally getting a little worse). A lot of that change for me is echoed in this post!

Every time I see people commenting on here saying "people still use that?" or "why would you do this to yourself" etc, I get it, I've been there, there's a lot of shit out there. But I stumbled on basically this path he describes (made a brand new account, stopped being self conscious, just started following interesting people and replying to interesting things) and I feel like it has supercharged my learning & connecting with cool people and maybe even "dark smoky rooms full of interesting people" (nothing smoky yet but getting there!)

OmarShehata··on Rabbit: LLM-First Mobile Phone
They largely use semantic metadata (like alt tags on images). This is part of why the "google docs is moving to canvas rendering"[1] caused a big stir there (accessibility would need to be implemented from scratch, possibly by using a hidden parallel DOM)

[1] https://news.ycombinator.com/item?id=27129858

OmarShehata··on Show HN: Auto Wiki – Turn your codebase into a Wiki
The Bitcoin and Mastadon links don't seem to be working! (wiki not found)

Would love to see this for Godot (https://github.com/godotengine/godot). Maybe Maplibre too (https://github.com/maplibre/maplibre-native)!

OmarShehata··on AI generated security reports about curl
Wow, how well did this land for you at the time, and others in your class? This is a very tricky thing, and I think most adults aren't equipped to deal with this.

(I think the average person is totally capable in telling these apart, the problem is that you kind of need to get used to reading things in a systematic way/put in some effort. It's very difficult to do this when scrolling through your phone late at night etc.)

OmarShehata··on Volkswagen Will Bring Back Physical Buttons in New Cars
This seems like an extreme position that at best wastes everybody's time and doesn't improve anything (and is a good example of why being careful with regulation like this is very important).

Consider settings like: how long it takes for the car to shut off the lights automatically after you turn it off and lock it. Or other obscure things. You're not going to have physical buttons for that. You'd have a digital screen with next/previous/ok buttons. That isn't any better than using a touch screen to access those settings. Arguably worse UX.

Another posted suggested allow touch screens but have them not operate while car is in motion, with no override, and I think that's a much more targeted/palatable proposal with (maybe?) less unexpected consequences.

OmarShehata··on Created a free website that simplifies stock analysis
It appears to be his startup idea, so I imagine they're looking for ways to monetize this but it's not clear what the plan is:

> I've been working on my startup idea for the past 2 months now. The idea is to provide data science tools to any retail investors regardless of their background. Everyone should have a fair chance to analyze the market and hopefully it makes the life of the user much easier.

8 months ago from: https://news.ycombinator.com/item?id=35222902

OmarShehata··on Things I learned from teaching
> Somehow I need to make assignments which thread the needle between being too hard to solve and too easy

The author is describing precisely what ever video game designer learns to do: build the right scaffolding at every step so people can be in Flow [1]

I've always enjoyed this symmetry between teaching & video game making. They both try to do the same thing in that way. (after all, video games are really just "voluntary work" when you think about it)

[1] See figure 1 https://www.gamedeveloper.com/design/cognitive-flow-the-psyc...

OmarShehata··on 15 years ago, I helped design Google Maps
> so it helps to have the spacial awareness to know that you're on the north west corner of Delancey facing north

Most of the people I've questioned on this struggle to figure out where the north west corner is or what direction north is etc. And yet they can still navigate in dense urban environments perfectly fine.

How? Landmark based navigation. All you need to do know is [some big landmark, like central park, or the sea] is behind you. Now to get to your destination, you want store X on your right and move forward. Turn right at store Y, etc. This is the reality of how a lot of people navigate and I think the new map reflects this.

OmarShehata··on LLMs cannot find reasoning errors, but can correct them
I think it does indeed suggest this, but I think this may be good news.

Part of what makes humans able to make progress in difficult, vague, and uncertain fields is a willingness to hold onto a point of view in the face of criticism to try & fix itl. This is, as a matter of fact, how science progresses, depending on if you ask scientists or historians of science. See Thomas Kuhn's Structure of Scientific Revolutions for more on this.

OmarShehata··on Your small imprecise ask is a big waste of their time
> there's no guarantee that you'll get what you want in that amount of time, and then what?

I am absolutely _shocked_ you've never received an answer to this.

There's no guarantee you'll get what you want in _any_ amount of time. You absolutely need to timebox, everything & anything, to make sure projects aren't going off the rails and ending up in the trash. This applies even to personal/solo work too. You can keep working on anything forever, it's much better to force a stopping point, get feedback (or ship it) and decide if it's worth continuing or not.

The mechanism at work here is forcing people to prioritize. If you're very good at prioritizing then a timebox maybe doesn't make sense, but even then, it may be hard to resist continuing to work on something to a higher degree of quality than is needed without a timebox.

OmarShehata··on First observation of a virus attaching to another virus
It's the same way chemicals have effects even though they have no energy/way to move themselves etc: they just diffuse in whatever medium they're in (like salt diffusing in water). No energy needed
OmarShehata··on Notes on making a multiplayer "doing nothing app"
Zenly (recently went bust) did this!

https://blog.mapbox.com/clear-the-fog-with-zenly-fc7778b06c3...

> Zenly’s location tech works in real time to uncover the map as you cover more ground. Zenly’s location tracking runs in the background, updating automatically as you check out new neighborhoods and take road trips.

(it's very social though I think, more "compete with friends!" than "explore your neighborhood & find delight in small things)

OmarShehata··on "Is Zero a Number?": Interviews with a Whole Class of Kindergartners
"I didn't know kids are that smart." Yes!! I wish more people recognized this. There's a "Free Range Kids"[0] movement growing out of this fact. A lot of people have ended up in a place where unstructured/unsupervised time is seen as a "waste" for kids, but I think this is the wrong direction. I think Khan Academy's new experimental high school[1] is also built around this fact (that kids are smart & just need access to resources & meaningful work)

[0] https://www.freerangekids.com/ [1] https://khanlabschool.org/About-Khan-Lab-School

OmarShehata··on Someone is trying to rebuild Friendster
This is a beautiful story, thank you for sharing! But also, I am so torn here.

On one hand, I want to say, it's worth making features like this even it sounds annoying to a lot of vocal people, because hey, maybe nudging people to connect more leads to some people connecting more and that was worth it!

On the other hand, maybe this is more a story of humans finding connection in spite of annoying UX, not necessarily because of it. These adventurous folks who were willing to connect on a DM on a whim online likely would have taken similar steps through mediums and find love (not necessarily each other)

OmarShehata··on This is what the year actually looks like (2018)
Highly encourage you to pose this question to friends, alongside "what do numbers look like in your head"

I keep learning new spatial relationships every time I ask! I drew mine here: https://omarshehata.me/notebook/my_number_forms

OmarShehata··on Shell using Fortnite, TikTok, etc. to convince kids fossil fuels are cool
Some people really do follow this line of thinking all the way down to "the industrial revolution was therefore bad for humanity", but I think most people would agree that not true, so these simple narratives must be incomplete

Jason Crawford has a really good analysis of this sentiment, trying to understand why people think this, on his Roots of Progress blog: https://rootsofprogress.org/the-spiritual-benefits-of-materi...

OmarShehata··on Security weaknesses of Copilot generated code in GitHub
That would be kind of wild. Imagine a world where whether your system was secure was just a matter of remembering to tell the AI agent "& also make it secure" before it writes your code.

(could be quite real!)

← PreviousPage 3 of 7Next →