jdonaldson

I’ve been willing to cut Microsoft some more slack in the Nadella era, but this reads as classic “embrace, extend, extinguish” all over again.

Homebrew starts from the MacOS CLI apps provided through XCode. It's a usermode package manager built on top of XCode components. MacPorts builds everything from scratch, which grants more power and optimization opportunities, but comes with many, many more rough configuration edges to deal with.

The optimization opportunities for some packages are typically on the order of 2x-5x faster, which isn't enough to deal with the headache of extra config for a personal dev box. So, my rule of thumb is, unless I know exactly the performance trade off I am trying to take advantage of by going with MacPorts, I will be better served by using Homebrew. I will wind up spending more time fixing config then I will have saved on execution processing time.

PG's articles are like a modern day Siddhartha. He illustrates the glories of growth and change initially, but by the end he illustrates the glories of stability and security.

I see that intellect is valued more and more intensely by a smaller and smaller group of people. I'm not sure what that means, but it gives me some small hope.

Yeah, I don't like this title. Coding for backprop is worth getting excited about, but please don't assume it supersedes all forms of "predictive coding". Plenty of predictive learning techniques do just fine without it, including our own brains.

In keeping with the No-Free-Lunch theorem, it's also highly desirable in general to have a variety of approaches at hand for solving certain predictive coding problems. Yes, this makes ML (as a field) cumbersome, but it also prevents us from painting ourselves into a corner.

Can you be more specific? Hofstadter specifically includes Cantor and Gödel as a way of showing how certain types of problems can't be solved with symbolic logic.

Agreed. He's a humble guy, but it's also very humbling being around him and listening him to talk. As I've gotten older, I have to look back at my STEM career in academia and industry, I have to say that he's the last individual that I've really been in awe of.

Maybe we recognize the contributions and influences of others in his work? Hofstadter is a big time guy, but he's not a self-centered egoist like Kurzweil. He'll regularly talk about the ideas he's bounced around with buddies like Daniel Dennett, etc. I see him as more of a poet than a philosopher or scientist. He is piecing together little bits of truth from many, many, different domains and showing how they reach the same conclusions. There's been nobody like him for over 50 years, so it's very difficult to draw comparisons to others, or even measure the impact of his contributions.

It's funny how behind the times that Wired is getting. Even my parents know about how scary good these text models are getting.

I might offer a slight wrinkle to this assertion. While raw power has driven an advancement, what Hofstadter is asserting is that symbolic reasoning (particularly recursion) is "unsolvable" for certain classes of problems. In other words, ML lives in a "box" that Hofstadter has defined. His work is still useful as a "lens" to understand the limits of AI, and what else it could possibly accomplish.

Hofstadter is a polymath, and he's arguably changed the understanding of the philosophy of existence as much as the understanding of these philosophies have changed him. I actually like using "they" to describe this process.

There is no single measure of intellect. We as a society just set some arbitrary goalposts and added some ways to measure contribution. The ability to take a step back and try to better understand reality will always be valuable, and there will always be a horizon there that we cannot measure.

I come away really disappointed with efforts for "meta search engines". There's so much more that should be possible. There should be a way to surface information regarding the results to that they can be combined (e.g. provide raw tf/idf scores, or expose the difference/bias between the results from different engines). I don't just want an API over triggering queries, I want an API that helps interpret/debias/rerank results.

Yeah, I'm sure that other plugin is great and all, but once Tim Pope puts his name on something, it becomes an instant default for me.

I suppose it depends on the definition of art and craftsmanship. Sometimes people tolerate lower reliability, and sometimes that’s ok for “art”. E.g. Harley Davidsons are not reliable, but that’s almost by design at some point. My German car mechanic was almost a social scene... owners were in and out so much we got familiar with each other and would go grab a beer while we waited for certain work to be finished. Point is, you don’t even mind that much if you really love the product. Italians and Germans just hit different with their industrial design. I fall in and out of love with each of them for different things... cars... espresso machines ... bikes. I don’t think the world would be a better place with only highly reliable products.

Italians have this long tradition of turning complex engineering into an art. Despite the fact that the Renaissance has come and gone, the roots for it run really deep in Italy still. I hope the same thing will be said about Silicon Valley one day.

what os/platform are you on?

I'm assuming you're familiar with tabloids?

https://en.wikipedia.org/wiki/New_York_Post

It's interesting to note that the original implementation for this was in Haxe, 6 years ago! https://news.ycombinator.com/item?id=8325700

All of what you said was true. Scientists try and find "original sources" for data, but they're usually analyzing measurements that have been altered in some way before they receive them... usually in the measurement hardware itself. One can't really avoid it completely. The flip side is that noise can actually be helpful, especially in situations where the signal itself is too weak to reach some detectable threshold on its own.

I generally try hard to remove bias/noise from data before analyzing it, but I also add noise back in later on... e.g. jittering data for plots, or putting a noise feature into a model to establish a threshold for useless features. I'm also a bit miffed that I had to discover this all on my own... Noise is useful and interesting! In business-oriented analysis, noise can be even more important, since changes in "data noise" can indicate important changes in business patterns, or problems with how they're collecting data (companies seldom share this information with analysts beforehand). I'd go so far as to say that understanding the noise in a dataset is the primary concern for most of my analyses these days.

Here's a good book with more pop-sci noise examples: https://www.amazon.com/Noise-Bart-Kosko/dp/0670034959

E2E group chats are closely related, but ephemeral. IPFS seems also closely related, but it's dealing more with the content of a file, and not really designed for handling the identity of who wrote or distributed the content.

Well, the scenario would involve distributing information where one couldn't/shouldn't rely on a central authority. Imagine using this to coordinate protests against a government, or in countries where there's too much corruption to trust any centralization at all.

These are just the details I gleaned from the article. But, I have a similar understanding... that the blockchain functionality is similar to self-signing.

The article bills Vapor as helping to decentralize the "data layer" of an api. This naturally has some immediate caveats... like... this sort of framework would not be suitable for real time or streaming Apis.

Also, the article doesn't really touch on the cost of maintaining all of this. It's true that vapor api hosts can charge a small transaction fee to maintain/run their processing nodes. I have to wonder what that looks like, and if people are willing to pay a per-request fee for an http transaction.

I don't believe this is intended to replace standard authentication, and calling out how it's impractical for common scenarios is fair game. However, I think it's more interesting to think about scenarios where this might be useful.

SSL certificates have their own issues, typically rooted in a centralized authorities dropping the ball or intentionally issuing fraudulent certificates

https://www.eff.org/deeplinks/2011/04/unqualified-names-ssl-...

https://security.googleblog.com/2013/12/further-improving-di...

Vapor also addresses decentralization in distribution of content as well.

I'm extremely skeptical of Vapor for other reasons, but we ought to at least keep the value proposition clear here.

* in some cases

Except scatter plots do not handle occlusion. If you have a lot of data to plot, the blob shape might be outliers.

Async is difficult to get right without proper generics and type parameterization support in the syntax. Brackets help collapse expressions to single lines, or expand them as needed. I like having that flexiblity.

Yeah, I'm surprised by these numbers. I think trains might end up being more efficient in an "ideal" sense. The rails really help remove friction from the equation. TIL.

I suppose you'd have to factor in other issues in the "real" world, like managing inclines, curves, and starts/stops for traffic. Boats generally have fewer constraints on those fronts. Also, boats can increase their size in more than one dimension, which lets them increase size more geometrically and efficiently (trains can only get longer, and hits practical length limitations faster).

I didn't know some of this, thanks! I used to live near an active train yard, and one of the warnings people gave me when I moved there was the sound caused by the trains "stacking" (which is the noise caused by each of the car couplings pulling out of their slack).

I thought it was a drawback at first, but now that I've moved away I kind of miss it.

Price aside, boats are a fairly compelling electric vehicle. Boats are one of the most efficient means of transporting large amounts of freight : https://en.wikipedia.org/wiki/Energy_efficiency_in_transport

In short, doubling the weight on a ship does not double the energy needs required for moving it. The same can't be said for wheeled transportation, let alone air. Batteries are very heavy, but on boats this doesn't have to be a big negative.

However, we're not talking large amounts of freight here, this is more or less a personal watercraft. Also, for personal watercraft most of the energy is spent on accelerating and decelerating, something that boats are not as efficient at.

So, good idea in theory, bad idea in application.