HNHacker News
TopNewBestAskShowJobs

mattnewport

3,559 karma · joined November 24, 2011

Cofounder / CTO at Osso VR.
submissionscomments
mattnewport··on 12% of Americans eat half the nation's beef
Now do some other foods and demonstrate that this isn't a typical distribution of consumption for any food item.

I guess 12% of Americans eat half the nation's lentils would not be as headline grabbing?

mattnewport··on Exploring the design space of binary search trees
Skimming through, I didn't see any discussion of AA trees which I have seen discussed as similar to red black trees but simpler to implement and have been interested in exploring for learning purposes. Any reason you didn't look specifically at AA trees, or were they covered somewhere that I missed?
mattnewport··on Owner gives up 2 big S.F. hotels expects city’s recovery to take up to 7 years
"reflected a plunge in bookings and a pandemic recovery expected to take far longer than expected."

Expected to take far longer than expected is a rather strange phrasing...

mattnewport··on Is This a Branch? (2021)
Code that uses min, max, clip, clamp, saturate, etc. when that is straightforwardly what operation is needed is clearer, more explicit, more concise and potentially faster. When used this way they are just straightforwardly better than equivalent if then statements.

Occasionally you see code that uses combinations of these in less obvious ways to do some more complex operation and I can see how one could argue that in rare cases that ends up less clear than if then statements but I've never really encountered that in many years of reading shader code.

There are other reasons to prefer functions over if then in shader code not discussed here as well, like replacing step with smoothstep for antialiasing.

mattnewport··on MiniZinc
I can recommend this course and its sequels on Coursera to anyone interested in learning MiniZinc: https://www.coursera.org/learn/basic-modeling

I enjoyed the courses and learning the techniques through the exercises but I still haven't really found a good application to a problem I need to solve for work or side projects.

mattnewport··on Ant Colony Optimization Algorithms
Are there applications where ant colony algorithms are actually widely used and have state of the art performance? In my experience with the travelling salesman problem they do very badly compared to simpler and faster methods like 2-opt.

It sounds like a neat idea but I haven't seen a problem where it works well in practice. I'm curious if there are real world success stories.

mattnewport··on Ask HN: How do you manage your personal finances?
We do "envelope budgeting" but taking advantage of a bank (Starling in the UK) that makes this cash free.

Starling lets you divide your account into spaces and you can have virtual debit cards associated with a space, or pay bills by direct debit straight from a space. At the beginning of the month the monthly budget for each space is put into that space and most spending comes directly from its space using a virtual card (in Google Wallet for physical purchases or with the appropriate card info for online purchases) or via direct debit. A few transactions are on the main card and are categorised later by transferring from the right space in the app.

Some budgets we expect to pretty much fully use each month, like groceries or bills. Others like the car budget we aim for rollover to cover occasional big expenses like a service. We also have an emergency fund to cover unanticipated bigger expenses.

I've tried various methods and software in the past and this has worked best. Tracking everything in an app or using ledger / beancount is too much work for too little benefit in my experience. It also hasn't worked well for me for budgeting as it tells you after the fact where your money went but doesn't by itself help / force you to stick to a budget the way the envelope method does.

mattnewport··on Bevy XPBD: A physics engine for the Bevy game engine
If you have a good understanding of modern C++ then I don't think you'll find Rust that hard to learn. I think most people talking about a steep learning curve are probably coming from garbage collected languages.

I've found Rust just formalises and gives the compiler the ability to check and enforce concepts already familiar to most C++ programmers confidante with move semantics, the lifetime of temporaries, etc.

Deciphering errors from the borrow checker take some practice still but for C++ programmers used to deciphering template instantiation errors it's not so hard.

mattnewport··on Ask HN: What's a not so famous software company that makes a lot of money
Maybe this doesn't fit your definition of "software company" but Renaissance Technologies certainly makes a lot of money: https://en.wikipedia.org/wiki/Renaissance_Technologies
mattnewport··on Ask HN: Best search engine alternatives to Google?
I've been paying for kagi for a while now and am pretty happy with it. Switched from DuckDuckGo.
mattnewport··on Cakelisp: A performance-oriented Lisp-like language
It's not a particularly controversial statement in the context of games. Many commercial games are still developed entirely or primarily in non garbage collected languages like C++.

Even Unity which uses garbage collected C# for game code is still a C++ engine under the hood for much of the performance critical engine code.

Having worked in games and VR development for many years I can say that memory management was not a major problem in C++ and for tricky areas of memory and resource management like GPU memory garbage collection doesn't generally help you.

When it comes time to optimize memory usage and performance it's often easier when memory management is explicit, deterministic and visible as well as highly customizable than when you have to work with a garbage collector which is a bit of a black box and has a lot of non deterministic and unpredictable behaviour.

Even after working for several years primarily in Unity with C# I still tend to prefer non garbage collected languages for games.

mattnewport··on Why Learn Compilers (2021)
I used C# for the first part, easy to translate from the Java code but you can make things more concise with modern C# features while still having access to a GC which is kinda necessary for the first part of the book.

I got rid of the code gen and visitor pattern stuff by using records in C# with pattern matching which felt a lot simpler to me.

mattnewport··on Can we create a thread for some of the best materials on CS available online?
Some Coursera courses I have found to be good:

- Stanford algorithms specialization: https://coursera.org/specializations/algorithms

- University of Melbourne Discrete Optimization: https://coursera.org/learn/discrete-optimization

- Basic Modeling for Discrete Optimization: https://coursera.org/learn/basic-modeling

- Advanced Modeling for Discrete Optimization: https://coursera.org/learn/advanced-modeling

Crafting Interpreters book mentioned elsewhere also very good.

mattnewport··on The Bronze Age has never looked stronger
I'm not sure if the Ruby Goldberg pun was deliberate or a happy autocorrect accident but I appreciate it either way!
mattnewport··on Testing GPT 4's code-writing capabilities with some real world problems
Is that obvious? The history of programming has been full of things that have at least claimed to increase programmer productivity, everything from high level languages to smart refactoring IDEs, debuggers, PaaS, etc and in all that time the trend has been towards more jobs for programmers not fewer.
mattnewport··on A Few Words About Bullshit
I couldn't disagree more. The thing being criticized is so much worse than the criticism here, I actually think the criticism is rather too measured.
mattnewport··on Open source is democratizing video game development
It is the id model, and it's not without value, it's just short of what I'd really call an "open source game" since so much of a game is about the assets, levels, etc. and not just the code.

I think id's open source code has been great for lots of reasons. I just think there is untapped potential in "fully" open source games where all the assets (including the source assets like Blender files and not just the intermediate formats like .fbx) are open source as well as the code as that makes it easier for people to experiment, learn and remix the content.

mattnewport··on Open source is democratizing video game development
The article mentions asset stores / packs as part of the democratization but that is also often the biggest barrier to open sourcing games. Solo or small indie developers often rely on asset stores for at least part of their content and that usually means they can't open source their entire game even if they wanted to as licenses for assets usually do not allow for redistribution in this form.

If you want to open source a game you generally need to plan for it from the start and only use assets that you fully own or that have licenses that allow redistribution.

You can open source just the code but this is limited in what it allows others to do in terms of contributing to or remixing the game. Even learning from the code is limited if you don't have access to the assets such that you can run the code and make changes.

mattnewport··on Coaching for “Normals”?
Are you sure about that?

"Steve Jobs had Bill Campbell as a mentor; Mark Zuckerberg had Steve Jobs; Bill Gates had Warren Buffett" - https://www.forbes.com/sites/brianrashid/2017/05/02/3-reason...

A quick search seems to confirm that both Steve Jobs and Bill Gates had mentors at various points in their careers.

mattnewport··on Coaching for “Normals”?
Sites like MentorCruise exist that offer paid mentoring with experienced people, including executives and startup founders. I'm not sure what a "relevant price point" is for you but they are often relatively affordable on a software engineer salary.

Full disclosure: I'm a startup founder with a profile on that site but there are other similar services out there I believe.

mattnewport··on Ask HN: How did you stop drinking?
What worked for me was Allen Carr's book Easy Way to Control Alcohol https://amzn.eu/d/89NkrKR

I tried several times to stop drinking previously, the longest I managed was for about 3 months. I think the hardest part is really deciding that you want to stop drinking. Once I did that, I read the book and it was easy. I haven't had a drink in about 6 years since.

What I like about this book is it doesn't make it into such a big thing like AA does. I don't think I was an alcoholic, though I did have some problem drinking types of patterns, and maybe the AA approach is necessary for full on alcoholics but I don't have to avoid going to bars or social situations where others drink for fear of relapsing, I just don't drink anymore.

I know this book has worked for a lot of people but no doubt it doesn't work for everybody. It made something that at times seemed impossible for me seem easy though.

mattnewport··on Far-ultraviolet LED designed to kill bacteria and viruses without harming humans
This might be useful but I wish they'd stop tying this sort of thing to covid when we now know it was never transmitted via surfaces and the society wide obsession with disinfecting everything was a complete waste of time for covid.
mattnewport··on Ask HN: Was it worth it for Go to add generics
People who care about generics already decided not to use Go because it didn't suit them. People who used Go value different things in a language or they wouldn't have been using Go.
mattnewport··on The end of classical Computer Science is coming
An AI capable of replacing human programmers will have to be a full AGI. At that point worrying about your career is probably moot.
mattnewport··on On AlphaTensor’s new matrix multiplication algorithms
Proofs are a pretty important part of algorithm development. Proofs of correctness as well as proofs of algorithmic complexity bounds. At the moment that doesn't seem like something this type of approach can do much for, though perhaps it could be combined with work on automated theorem proving.
mattnewport··on Cloud desktops aren't as good as you'd think
Game and VR development generally need powerful desktops both for programmers and content creators. Our entire team of (fully remote) VR developers have high end gaming desktops. If cloud workstations worked well they'd be interesting for us but none of them really support VR development at the moment.
mattnewport··on Values Are Facts
There seems to be a bit of a sleight of hand going on here. The argument being made is mostly about what values people hold being facts. Usually when people make a distinction between facts and values though they are not saying that what values people hold are not facts but rather that what values people should hold are not facts. Is vs ought.

Obviously the author understands this distinction so I'm not sure why it seems to be ignored / glossed over here.

You can make a case that what values ought to be is also connected to facts and I would tend to agree with that argument but it's not really the case that is being made here.

mattnewport··on Stadia Is a Major Driver of Vulkan Adoption
It seems to be a Stadia fan site so the reason for this article existing is that they need to keep a steady stream of Stadia news and puff pieces. They're just trying to come up with positive things to say about Stadia, not trying to actually come up with true insights about the broader industry.
mattnewport··on The era of fast, cheap genome sequencing is here
And yet the era of being able to do anything really useful with the information has yet to arrive.
mattnewport··on DreamFusion: Text-to-3D using 2D Diffusion
I'm saying something different, that the most impressive examples of AI breakthroughs are doing things that humans find hard / are bad at. Meanwhile there are many things that people find easy / do without thinking / can be done by dogs or very young children that AI struggles with.

It suggests to me that what most current approaches are doing is something fairly different from what human / animal intelligence is doing in important ways. That means we will likely continue to see AI do increasingly amazing things while at the same time struggling to perform a lot of tasks that are quite basic for humans.

It is the fact that AI is proving to be a better artist than most humans while not being able to do many things that are simple for a 4 year old that suggests strongly to me that some of the fundamental mechanisms are fairly different still, or current AI approaches are missing some key insights.

I could be wrong. I'd bet money that I'm right if there was an easy way to do it though.

← PreviousPage 2 of 29Next →