HNHacker News
TopNewBestAskShowJobs

goldenshale

411 karma · joined December 25, 2015

submissionscomments
goldenshale··on Pause Giant AI Experiments: An Open Letter
How silly. Imagine telling Elon to pause development of all new rockets until people become comfortable with Starlink? Are language models really that scary?

It feels like it has become trendy for everyone to worry about everything rather than to look to the future, be creative, and explore all of the amazing uses and problems that we can start tackling.

goldenshale··on We’ve filed a law­suit chal­leng­ing Sta­ble Dif­fu­sion
Lame. These are scared opportunists who are going to screw things up for everyone because they are afraid of the unknown. Seriously, do something new and creative if you are an artist. Design your way into relevance.
goldenshale··on NanoGPT
Bad ass! A great addition would be some content on tuning pre-trained language models for particular purposes. It would be great to have examples of things like tuning a GPT model trained on language and code to take in a context and spit out code in my custom API, or using my internal terminology. Not sure if this is RL based fine tuning or just a bunch of language to code examples in a fine tuning dataset? In essence, how can we start using language to control our software?
goldenshale··on High-documentation, low-meeting work culture
If only we could have an AI do the documentation...
goldenshale··on Greg Bear has died
Yes, Eon! Amazing book. Darwin's Radio is a really fun idea too. Definitely recommend that if you enjoyed Eon. Very different, but equally creative and fascinating.
goldenshale··on Zig, the Small Language
Also shouldn't the input argument a be named items instead? From what I can see items is an invalid variable as is.
goldenshale··on Ask HN: What Are You Learning?
- Quantum computing via https://quantum.country/

- Cooking via YouTube (Food Wishes w/ Chef John, and Kenji Lopez Alt's channel)

- ML model architectures (AI Coffee Break, and Yannic Kilcher's channels)

- Exploring different ways of thinking about and responses to the question: What is the meaning of life? (I don't think there is one, so it's more about how we derive one for ourselves.)

goldenshale··on We will never have enough software developers (2020)
I think this article hits on some truths and gets a handful of things wrong. First, they are correct that we are in a dynamic profession that requires constant learning and expanding. No doubt the people who choose to stay in software are likely to be people who are curious, life-long learners where this is a benefit of the profession rather than a drawback. That said, one thing I noticed from teaching computer science while a graduate student is that the poor students think about languages and libraries as key skills, while the better students think about the data structures, algorithms, and design principles. These slightly more meta-level concepts are far more stable and timeless, while the underlying implementations of them are constantly evolving. If you think of each new library as "having to learn a new skill" I can imagine burnout and overload are more likely, but if you figure once you know 3D graphics you can pretty easily pickup any new graphics engine, then it might even seem fun to explore different takes on the same old concepts. It's like hearing a new band in your favorite genre.

As for attributing burnout as the core issue here, I would strongly disagree with this idea. When teaching undergrads I noticed immediately that a good portion of each student cohort was basically there because they were interested in making money in the future rather than exploring the ideas of computer science. They were no doubt going to get frustrated or bored and move into management or some other profession rather than continue to expand as an engineer. This is totally fine, and they are probably richer for having learned what they did, but I don't know why we can't just see this and appreciate it for what it is rather than portraying it as the drama of burnout.

goldenshale··on Why do you waste so much time on the internet?
I wonder if this is why the internet hasn't led to the dramatic increase in productivity that one would have predicted a global communications network would have produced? Instead we are turning ourselves into mice with a dopamine feed on tap.

Maybe we need to find ways to value creation far more than consumption. Missing new and cool ideas is peanuts in comparison to creating something from nothing, whether that be art, technology, or friendship.

goldenshale··on Starting Your Computer Music Journey with Clojure and Overtone in Emacs
Actually the instruments that come with Sonic Pi are created in Overtone. Sam Aaron and I were expats and best friends living in Amsterdam, and we created Overtone together. He then went on to create SonicPi, and I joined DeepMind as the first employee.
goldenshale··on Developers of small modular reactors hope their time has come
Seems like a perfect pairing for direct air carbon capture systems, which are primarily energy limited.
goldenshale··on Russian forces invade Ukraine after Putin orders attack
The period of US hegemony from WW2 until today has by far been the most peaceful time in the history of civilization. Unlike previous super powers, the US did not take over any other countries or colonize by force. We have done many horrible things, like support dictators, invade Iraq based on lies (to depose a dictator), etc., but we have also supported people all over after natural disasters, helped with foreign aid to relieve famines, policed global trade to enable fair markets, and many other good deeds. It's definitely not the only way, and we could do much, much better, but show me another major country that you would prefer hold the reigns of global power?
goldenshale··on Times are great for programmers now. How does it end?
If you look at the data, software jobs are growing far faster than new graduates in computer science. A quick google search just found well over 250,000 open positions in the US, with about 50,000 new graduates per year. I'd bet we have decades more of high demand and high salaries for people with CS degrees. I wouldn't be surprised if low-code tools get good enough to power many websites though, so I wouldn't be quite as bullish on code-camp graduates seeing the same growth in salaries.
goldenshale··on Ask HN: What's the best way to monetize actual programming?
Find another company to work for. Not making people look bad for being less productive is absurd, and there will be many high performance teams where that is not an issue. Shoot for more prestigious companies, and there will be rewards for productivity. Either that, or find an early stage startup where you can help set the tone and culture.
goldenshale··on Ask HN: How do you get started as independent consultant or contractor
Your best strategy no doubt depends on who you are and who you want to be. Do you want to work on a wide range of projects, or stay focused on a niche, like embedded systems for hardware control, audio and music DSP, 3D graphics, games, web sites, ops and infrastructure, machine learning, algorithms for distributed systems, ...? You'll want to build a case for your capabilities and experience through a portfolio of school projects, open source contributions, demos, etc. In your niche area or programming language eco-system, there are likely to be slack channels, news groups, mailing lists, and blogs, so become a part of a community, helping when people ask questions, fixing bugs people care about, or blogging examples and common pitfalls.

Imagine you are working at a games company trying to hire a contractor to help with 3D audio programming. The ideal candidate is probably going to have a portfolio that includes 3D audio programming, or at least audio and DSP experience.

Personally, I'd focus on building valuable expertise in a community that includes the kinds of people and businesses for whom you would want to work. There is always demand for skilled engineers who deliver.

goldenshale··on Content Addressable Memory and Neural Networks
This idea relates to one of the key results in neural networks, where it has been shown that a sufficiently large single hidden layer network can approximate any function. (In essence a bunch of piece wise segments that can approximate the function along an interval.)
goldenshale··on Why Co–Star Uses Haskell
Haskell: recommended by the insight of human astrologers.

There’s a poetry in the merging of two new-age religions stemming from different communities. Unfortunately I’m an atheist, and so are most of us here I’d imagine.

goldenshale··on Geothermal's path to relevance: cheap drilling
Not really. A ground-source heat pump uses the thermally stable ground below a house to heat in the winter and cool in the summer, while a geothermal power plant pumps cool water down a well and then gets hot steam out another connected well to drive turbines.
goldenshale··on Geothermal's path to relevance: cheap drilling
Yeah, per-house is probably never going to make sense, but larger scale operations where they can use hydraulic fracturing techniques to expand the surface area could be the solution. They call these Enhanced Geothermal Systems, and basically they frack the rock between two well bores to try to maximize connectivity and surface area.
goldenshale··on Evidence for European presence in the Americas in AD 1021
I highly recommend you read (or listen, the audiobook was great) to Conquistador by Buddy Levy. It might change your mind on this point. It tells the story of Cortez from the day they arrive on the shores of Mexico to when they take over the Aztec empire, where there were still many, many native peoples. The European advantages were many more than just disease... for better and worse.
goldenshale··on Cannabis and Male Fertility: A Systematic Review
Yeah, when we wanted to have a kid I quit cannabis and hot-tubbing for 6 months in anticipation of conceiving. It worked, and she's healthy! Sample size of 1, but it felt like kind of a nice way to commit to a family and get in the proper head space anyways.
goldenshale··on People in present-day Austria drank beer and ate blue cheese 2,700 years ago
They've also discovered neolithic villages that were along the shores of lakes in the alps in present day Switzerland that have been dated to as far back as 1,000 BC. They were already living in post and beam homes on stilts, herding animals, making pottery, and probably living a pretty nice life.
goldenshale··on Lab-grown meat may never be cost-competitive enough to displace traditional meat
Wait, so you mean if you take linear projections of today's outputs and costs and extrapolate into the future it seems impossible? No shit. This is basically the case with every interesting technology project. There will no doubt have to be innovations in the approach, the equipment, etc. But in reality all we are talking about is mixing fluids and maintaining some temperatures. All of which is done at massive scale for chemicals, oil and gas, etc. I don't have strong claims to make regarding cultivated meat, but the kind of thinking behind this article is rampant. If the money is there, then this can no doubt be scaled up for far, far cheaper than the author envisions.
goldenshale··on Organic and regenerative agriculture are revitalizing rural Montana economies
I'm a software engineer with a passion for botany and growing vegetables. I mostly attribute this interest to growing up with a gardening mother, but I think the mixture of engineering systems, working with equipment, and trying to balance complex trade-offs all lend themselves to the kind of person who likes to create virtual worlds in their mind. Writing software is like creating a little ecosystem of moving parts and interlocking components, and taming complexity is the eternal battle as systems grow. Many of us are probably drawn to analogous systems and problem spaces. Hacker News is basically a place where we share links, nerd out on ideas, and make snarky comments.

I think part of what is exciting and intriguing in your article is that it paints a picture of what agriculture could look like. Instead of just feeling a sense of dread about how our society is ruining everything, many of us are technological optimists with a core belief that science and engineering can help light a way to a better future. Who doesn't like reading articles that reinforce their existing beliefs? ;-)

I've scoured YouTube and other sources for regenerative ag content over the last few years, and what I crave are more specifics and details of business models, scientific analysis, etc. It would be fascinating to actually see the books for a couple of different farms, conventional vs organic vs regenerative. How much money is spent on chemicals vs labor vs seed, etc.? What does insurance cost, and how much risk is there? Does regenerative ag with higher soil carbon lower this risk? (Maybe there is a smart regenerative only insurance play here?) Many of us would probably love to work on autonomous tractors, computer vision to spot problems, etc., etc., so learning about the key levers and problems underlying this industry would be incredibly thought provoking. While reading your article my head went to: why can't those electric fences both monitor the cattle and move themselves around? Could you attach water or compost to the cows themselves to have them help distribute useful inputs? There is an incredible amount of food waste, yard trimming waste, and other carbon rich waste streams generated by society. Why aren't we spreading that on these farms? Instead of tilling the top inch or two, couldn't we be planting the seeds in a new layer of added material? I guess this would require moving a lot of mass, but maybe if it's an autonomous supply chain it could be constantly bringing beneficial waste streams to farmers? It seems like each crop is likely to take some nutrients and provide others. Could you not auto-sample a farm and then automically determine the optimal crop mix which will most benefit the soil? If you have market data you could also connect this with current market prices. Last, if all you did was grow optimal cover crops for a few years could you dramatically improve a farm? Should we pay farmers to give the land a break and only grow beneficial crops so they can come back in N years as an organic, regenerative operation?

My wife and I have started trying to order all of our meats (fish, pork, beef, chicken) from sustainable operations, and it's actually quite hard to know who to go to. When we do finally find one that seems compelling, they are often sold out. Is anyone creating a centralized marketplace for the output from regenerative farms? I have a feeling we aren't the only ones who want to support these operations.

Thanks for a thought provoking article!

goldenshale··on Beej's Guide to Network Programming (1994-2020)
What a classic! Back in 1999 I wrote the networking infrastructure for a distributed, evolutionary, fractal art system for Steven Rooke (maybe you remember the fractal absolut vodka bottle on the back of Wired magazine) and I learned everything from Beej's guide. This brought back great memories. What a serious internet classic for so many software devs. Thanks Beej!
goldenshale··on Amazon, Apple and Google Cut Off Parler
It's a free market as well as a land of free speech, so they can just find another hosting provider, or build their own servers. They can enable whatever kind of groups they choose, but that should in no way require other people or organizations to collaborate with them.
goldenshale··on Computational Model Reveals How the Brain Manages Short-Term Memories
My intuitive understanding of these findings is that short term memory works by maintaining a consistent firing pattern in a particular region of the prefrontal cortex. By default a given firing pattern might activate other patterns because neurons tend to be a part of multiple potential pattern groups, and so left unchecked it could result in an unstable, random walk type of sequence. To stay in the attractor of one memory though, inhibitory neurons are used to turn off everything that isn't a part of the memory, thereby allowing it to remain in a more steady state.
goldenshale··on Dark’s new backend will be in F#
Or maybe this is why Dark is a bad business endeavor in the first place. I would never bet my company's future on something like this, and the toolchain doesn't seem sufficiently powerful to warrant the risk.

Sorry, but this is no Clojure. Instead of a proper comment syntax you define a dummy variable to a string? Ghetto.

goldenshale··on Ask HN: How do you overcome decision fatigue in software development?
This is the art of being an effective person, let alone doing good engineering, and personally I find building and using this craftsman's intuition to be the great joy of working in software. Contrary to many of the comments here, I would never just go for a "boring framework" because it is tried and true, or just use what I already know. That sounds like a great way to grind yourself into burnout.

Instead I would start jamming on a whiteboard or in a notebook. Think about the problem you are tackling: what are the inputs, what are the outputs, what are the key functions, and what data structures and algorithms might be important? How about the team? Is it just you, or are you working with other people? What are your experiences, interests and passions? What's the timeline? Spend some time exploring libraries, reading paper and blog posts to inform yourself of the space. Maybe read some open source libraries. Don't go nuts, but scan around to soak up some good ideas and build a mental model of the space. This all helps put some parameters around the situation and give you some confidence that you are making informed decisions. Now you can think about the MVP, key interfaces, etc, and you might find that you can do some quick and dirty components for 80% of the project while really focusing on the heart of the challenge in that last 20%.

Often times if you define good interfaces you can swap early hacks and experiments out for something fancier (faster, more scalable, etc) later on, so then you don't need to worry as much about making optimal decisions. For example, you might use some distributed key value store or message bus in the end system, but to get the ball rolling you can just spin up a Redis instance during development, or even just write a couple functions that fake it in-memory or using JSON on disk for starters.

Now you can think about what you want to learn in this project. What's going to be fun and interesting about it? Do you want to dig into some new algorithms, a different language, a machine learning model? I like to use my curiosity as a major motivation to differentiate from other solutions and push through what could otherwise be boring aspects of the project. Say you need a service that just manages logins and sends back some tokens. That can be done in a million ways, and it is likely to get swapped out down the line anyways as you scale or whatever. So make it interesting, pick a new library or language and go nuts. This of course needs to be balanced with making sure you can accomplish the task in a reasonable amount of time, but this kind of learning is empowering and it will make you a more capable engineer. Now 20 years into my career I can program in 10+ languages on embedded micro-controllers or virtual cloud clusters, and picking up a new language or framework isn't daunting at all.

Have fun doing it, and you'll get better faster than everyone who is picking the boring framework in their same old language to solve it the way that the blogs are telling them to and then frantically searching stackoverflow because half way into the project they discover that their square problem doesn't perfectly fit into the framework's glorious round, ahem, hole. That's not engineering or craftsmanship. That's nervous fanboy frenzy that is the opposite of empowering.

Everyone has to find their own explore vs exploit trade-off, and we need people whose experience is narrow and deep as well as others who have learned far and wide. Choose what kind of person you want to be and make it happen.

goldenshale··on Amazon Interactive Video Service – Add Live Video to Your Apps and Websites
Because most recordings are never actually streamed to that many people. My guess is that the popular and expensive streams make profit while the rest cost basically nothing to host.
← PreviousPage 2 of 3Next →