HNHacker News
TopNewBestAskShowJobs

romaniv

2,750 karma · joined February 2, 2011

submissionscomments
romaniv··on CRPG Book Released
It's a finished product that just needs a physical release. Kickstarting it would make a lot of sense.
romaniv··on Ted Nelson on What Modern Programmers Can Learn from the Past [video]
Generally, I'm wondering the same thing. It's not all bad, though. Alan Kay has his VPRI, which has a lot of talented people.

Bret Victor has some good conceptual stuff. But as he said himself:

https://twitter.com/worrydream/status/947177766885081088

"As Kay himself has said so many times, this question is entirely about funding. With ARPA/IPTO-style funding, you get Kays (and Engelbarts, Minskys, McCarthys, Corbatós, etc) With VC and NSF, you don't. Even Kay hasn't been able to "be Kay" since the 80s: http://worrydream.com/2017-12-30-alan/ "

Another problem, I think, is that many people are no longer interested in this kind of research. Two popular lines of thought are:

- Our infrastructural computing is already awesome or at least good enough, so we do not need anything simpler or different, just more tooling and automation.

- Machine learning will fix everything in computing.

romaniv··on Ted Nelson on What Modern Programmers Can Learn from the Past [video]
I am constantly amazed by the productivity of top computing experts in the 60s and 70s. They managed to accomplish so much in such a short time, while working in small teams, on ridiculously slow hardware and having next to no libraries or tooling.

In the corporate world today it often takes a dozen people and a few years to create what's effectively a website that records user inputs to a database and occasionally sends emails.

I understand that I'm comparing top programmers of the past to average programmers today, but I think modern tooling and accidental complexity of our computing environments plays a large part here as well.

romaniv··on Fuze for Nintendo Switch: An easy-to-learn text-based programming language
Depends on what you want.

If you want to learn about functional programming, something like this could give you fun and relatively painless head start: http://www.4clojure.com/ . Also, you could try to download LINQPad and look at code examples.

If you want to learn about real OOP, I would recommend looking into Smalltalk, learning about design principles behind Xerox Alto/Star and listening to some OOP seminars from mid-80s.

If you want to learn about high-level system design, here are some intelligent people I would listen to: Alan Kay, Rich Hickey, Peter Norvig, Salvatore Sanfilippo.

I would also recommend every programmer to watch through at least one introductory AI class. Not because they necessarily need to work with AI algorithms, but because those classes teach to solve programming problems in a better way than just "write more logic and abstractions".

https://ocw.mit.edu/courses/electrical-engineering-and-compu...

https://www.youtube.com/watch?v=mkBM4jvrhzk&list=PLAwxTw4SYa...

Hope this helps.

romaniv··on The Death of Microservice Madness in 2018
>Can we extend that logic to classes or interfaces?

This was one of the main ideas behind the original definition of OOP. The original notion of "object" was very similar to our current notion of "service":

https://www.youtube.com/watch?v=QjJaFG63Hlo

Objects received messages, including messages sent over the network. There was not supposed to be a clear distinction between local and remote services - by design. A lot of inter-computer stuff could be/was handled transparently.

romaniv··on Fuze for Nintendo Switch: An easy-to-learn text-based programming language
First, we teach people variables, loops and conditions. Then we teach them procedures. Then we are horrified that newly minted "programmers" write code that is a giant mess of nested loops and conditionals, split across arbitrarily created procedures that manage bloated and fully mutable program state.

Mental models are important. Life would be better for everyone if people started learning programming with LISP, Smalltalk, or at least C#'s LINQ in LINQPad.

romaniv··on No, machines can’t read better than humans
Scientists often carefully construct tasks, evaluation criteria and even test sets to make AI seem better than humans or at least better than it really is.
romaniv··on Neural Style Transfer for Musical Melodies
Modern step sequencers (for example, Elektron boxes) are way more than just a grid. They have microtiming, parametrized triggers, parameter sliding, probabilistic and conditional triggers, and are capable of running multiple patterns of varying lengths that reset at different rates.
romaniv··on Neural Style Transfer for Musical Melodies
>- I haven't heard the term transposition for samples - is this the name for moving samples in pitch and compensating in time so that it sounds undistorted?

Yes, that's exactly it. There are traditional algorithms that do it, but most are variations on time stretching and granular synthesis. They aren't as good as they could be. A good algorithm would be able to take in a handful of examples and generate a model that seamlessly interpolates across different pitches and velocities, accounting for differences in timbre, attack, decay, noises and so on.

Thank you for the links.

romaniv··on Neural Style Transfer for Musical Melodies
Well, yes, that's how convolution reverbs work. Except I've never seen anything that would allow users to record a reverb with conventional hardware and without any elaborate setups.

Also, there is more to reverb than just impulse response. A more sophisticated reverb simulator would map out the space and allow you to choose where the "listener" and the source are located in that environment.

The nearby comment about moving obstructions is on-point as well.

There is ample space for application of machine learning here.

romaniv··on Neural Style Transfer for Musical Melodies
Generating rhythms is a problem that has been solved by arpeggiators, step sequencers, analog modular rigs and more sophisticated tools like KARMA. You don't need machine learning for it.
romaniv··on Neural Style Transfer for Musical Melodies
>Why do we need Style Transfer for Music?

We don't. Most of the applications of ANNs to music that I've seen so far are solving problems that either don't exist or have been very successfully solved by much simpler methods with much better results.

If you want to apply AI to something useful in music, how about:

- Machine learning for reverb generation. I.e. being able to "record" reverb from real space and apply it to sound.

- AI/ML for high-quality sample transposition.

- AI for pure sound synthesis. Imagine something that does PCA a set of sounds and then allows you to generate new sounds by changing components with a bunch of knobs.

- AI for real-time sample morphing. For example, being able to synthesize a choir that sings the words you need. Or, how about transforming your own voice into a quality choir vocals? Hey, style transfer!

- AI for high-quality music transcription (sound to notes or MIDI). This is being done to some extent, but I don't think it's very good yet.

And so on.

I can assure you, every singe one of these would be something musicians would be willing to pay good money for.

romaniv··on Turning web design mockups into code with Deep Learning
>Currently, the largest barrier to automating front-end development is computing power.

Actually, it's poorly though-out web technologies that don't allow for reasonable WYSIWYG editing.

romaniv··on A letter about Google AMP
>Fast websites stayed fast, and slow websites stayed slow.

Website did ridiculous things for SEO when Google was the main traffic-driver on the web. Ridiculous. If Google sufficiently penalized bloat in ranking most websites would remove the bloat. But it never was a significant enough factor.

romaniv··on Rubbish (2002)
When someone says they don't mind surveillance and have "nothing to hide", ask them for the password to their primary email. Most people simply don't think through the practical implications of their information being sifted through.
romaniv··on AI in drug discovery is overhyped: examples from AstraZeneca, Harvard, Stanford
I'd say this is more of a result of decades of trial and error in different architectures/topologies rather than some fundamental property of neural networks in general.
romaniv··on Open source modular synth software lets you create 70s & 80s electronic music
There are tons of different workflows to create electronic music. You can record sounds (and different instruments) linearly in layers. You can play a single keyboard set up in some clever way that allows you to trigger and control multiple virtual instruments at once (often with the help of arpeggiators or something like KARMA). You can use a looper. You can set up a step sequencer that plays notes in a loop while you manipulate both sounds and melody in some way in real-time. You can set up some gear to procedurally trigger sounds, which is what modular racks are often used for (look up "krell patch"). Or you can do several of these things at once, possibly in a group.
romaniv··on Open source modular synth software lets you create 70s & 80s electronic music
> The edge for vintage analogue synths was the fact that they offer more hand on control but with the wealth of midi controls that come with a wide array of presets support many software synths its still not much of a good reason to pick a hardware synth.

There are at least two fundamentally flawed assumptions behind your post.

1. Electronic music isn't just about endless options and capabilities. Endless options give you endless search space to navigate.

2. The idea that generic computers are always better than dedicated hardware is rather archaic, because with modern technology you can integrate powerful computers into dedicated hardware very cheaply.

There is a huge difference between using a fully integrated, though-out hardware synthesizer versus using a MIDI controller mapped to some plugin in a DAW. There is a reason thousands of musicians still buy analog gear, VA synths and fully digital FX pedals.

romaniv··on Meltdown and Spectre
Maybe it's a sign that the current CPU architectures are too complex for their own good. Branch prediction is a really, really weird concept. It could be avoided if we switched from Von Neumann architecture.
romaniv··on A pattern language for microservices
There is a difference between designing a system where objects always behave similar to network nodes and designing a cludge... I mean, a protocol that allows procedures to be called across a network.

I'm not saying that message-passing is magic and transparently solves all the problems associated with distributed computing, but conceptually it's different from RPC. In fact, it's kind of the opposite of RPC. Keep in mind that Smalltalk 72 was developed in the same environment (Xerox Park labs) that gave us Ethernet and developed large parts of the conceptual framework underpinning our modern networking.

romaniv··on A pattern language for microservices
Interestingly, it seems RPC was one of those older ideas that was resurrected and replaced remote message passing when Smalltalk was phased out.

Alan Kay on RPC: "The people who liked objects as non-data were smaller in number, and included myself, Carl Hewitt, Dave Reed and a few others – pretty much all of this group were from the ARPA community and were involved in one way or another with the design of ARPAnet->Internet in which the basic unit of computation was a whole computer. But just to show how stubbornly an idea can hang on, all through the seventies and eighties, there were many people who tried to get by with “Remote Procedure Call” instead of thinking about objects and messages. Sic transit gloria mundi."

Source: http://mythz.servicestack.net/blog/2013/02/27/the-deep-insig...

romaniv··on A pattern language for microservices
I'm watching some OOP videos from mid-eighties right now. Can't shake the feeling that the original notion of OOP was solving the same problems as microservices try to solve today, but with far more elegance and more coherent theoretical framework.

(Example: https://www.youtube.com/watch?v=QjJaFG63Hlo)

romaniv··on A plan to rescue the Web from the Internet
Just because Kay criticized the Web for its limits on user participation does not mean he did not criticize it for other reasons.

Here is an entire talk on complexity and robustness in software where the Web is mentioned as one of the worst offenders of design by accretion:

https://youtu.be/QboI_1WJUlM?t=1046

(I'm linking to a specific part, but the entire thing is worth listening to.)

Also, he does talk about protocols. Heck, VPRI implemented a TCP/IP implementation by treating it as a language and using a non-deterministic parser.

http://www.vpri.org/pdf/tr2007008_steps.pdf

Page 17.

romaniv··on A plan to rescue the Web from the Internet
Having watched nearly every talk/interview of Kay I could find on YouTube (which includes the OOPSLA keynote) and having read some of his writings I think I have a reasonably good understanding of what he is advocating for.

Here is Kay talking about Internet's resilience in a bit more detail:

https://youtu.be/NdSD07U5uBs?t=1472

>His position is strongly aligned with the "mechanism, not policy" philosophy.

I don't see how this opposes or contradicts any of what I wrote above.

romaniv··on A plan to rescue the Web from the Internet
Saying that WebDAV lost to Facebook is a bit like saying HTTP POST and DELETE actions lost to Facebook.

I was thinking of publishing and versioning of a different kind. For example, right now the only way to see how some website looked 10 years ago is to go to webarchive.org. 5% of all links on the web get broken every year. Isn't that something worth solving more systematically? In this respect content-addressable Web is a very interesting idea.

romaniv··on A plan to rescue the Web from the Internet
The reason the Web needs rescuing is because it's not a particularly well-designed system that has been patched over and over again for the last quarter century. And now it has been degraded to a delivery layer for JavaScript apps, a poor RPC protocol and an overly complex UI rendering toolkit.

It should have had protocol-level technologies for preserving historic data, for search, for offline use, for authoring and publishing. If you look closely, cloud services created tools to easily do all those things and that's how they got in control.

"The Internet was done so well that most people think of it as a natural resource like the Pacific Ocean, rather than something that was man-made. When was the last time a technology with a scale like that was so error-free? The Web, in comparison, is a joke. The Web was done by amateurs." -- Alan Kay.

A lot of web devs were enraged by his comment without listening to the context. He was talking about the lack of protocol-level solutions for resilience of the Web.

romaniv··on Norway switches off FM radio, but a station is defying government order
Love it. It's one of the most well-designed electronic devices I ever owned. Rally fun to use. I think it's worth studying as an example of great user experience and designing for "emergent" features - even if you don't care about synthesizers all that much.

Also, OP-1's synthesis capabilities are much deeper than it would seem from reading the manual or watching a typical review. And it has an amazing online community.

romaniv··on AlphaGo Teach – Find new and creative ways of playing Go
This doesn't make sense. Computers do stuff that people don't "comprehend" all the time. In fact, in the past, one of the fundamental appeals and drives behind AI was to make systems that could explain themselves, not requiring complex analysis and experiments to understand what's going on. (And vice vera, systems that could understand humans without programming.)
romaniv··on Norway switches off FM radio, but a station is defying government order
Radios are still sold in stores. Also, Phones and MP3 players often have FM radios. Heck, my best radio is probably my OP-1 synthesizer.

The real "magic" of FM radio that it something that Just Works. You don't have to pay, you don't have to connect to some network, and it's something that worked since late 1930s. No matter what people say, any current digital radio formal will almost certainly obsolete in less than 20 years.

romaniv··on Google's AlphaZero Beats Stockfish In 100-Game Match
> Like, how does AlphaZero optimize for “how to slow Climate Change”?

I think that's a bit ambitious. GAI is not obliged to solve all the world's problems. But I would ask a more modest question: how could AlphaZero approach problems like web service integration? These problems are definitely solvable, but converting them to optimization problems is the really difficult part that requires genuine intelligence.

← PreviousPage 7 of 30Next →