tylerneylon

I fully agree. The US's academic system provides a lot of insulation to researchers because generally this publishing cost will be paid for by your employing company, by your university, or by a grant. So researchers are not incentivized to spend a ton of time worrying about it. At the same time, I believe the large majority of researchers don't realize how little (or negative, arguably) value they receive from paying for publication vs doing so for free (such as on arxiv).

Specifically, online academic publishing is, at its core, indexing and hosting pdf files. It is some work to do a good job. But it's also quite achievable to re-create the same service without asking for much, if anything, from authors. Given a little funding, every field could use arxiv or their version of arxiv (which is free to publish on). The bottleneck to a large-scale change is the self-sustaining prestige of a paid journal's badge.

As a first step, we can spread awareness among authors of how crazy it is to pay so much to publish.

I have lots of help. I write during work hours (so kids are at school). At work I delegate as much as possible to a team of great people; hiring a good team has been critical. I sometimes hire a sitter to support some exercise time, and other times I bring my kids with me to a running track. I hire housecleaners to clean regularly to help me keep my home nice and organized. I don't currently have a personal assistant, but in the past I did and that was helpful for getting "life things" done. My kids and I have gotten used to a culture where we alternate between independent time (so I can, say, read a textbook for an hour while they play) and together time. I've found that my kids enjoy independent time when they know they will regularly have focused attention from me, so I try to give them that (reliable focused attention, ideally 1:1). I do have less time to work than I did before kids, so I spend a portion of my energy planning ahead or delegating.

I'm in my 40s and feel sharp, with some evidence to back that up (such as career success and learning new things well). I believe these things help me stay sharp: I prioritize sleep. I try to get exercise (most often running) every day. I try to write every day; I write about ideas (typically new ideas in math or philosophy). I often read something I'm learning, such as a math book, something for work, or learning a new language. I meet weekly (online) with someone who's teaching me a new language - in this case French, but I think what's important is to keep learning. My work presents new technical challenges regularly, and I made a life decision to change my work whenever I feel burnout approaching.

To address other comments: I have two young kids, and I've been a single dad since my youngest was 3. Some commenters speak as if having kids kills your work or thinking life. This is partially true for the first year because of sleep depravation. And it's a new level of difficulty in terms of organizational demand on your life. It's harder. But it doesn't kill your mental abilities, and in fact I'm a happier person with my kids around, which gives me more motivation and life satisfaction. I've been able to solo run a successful bootstrapped startup while being a single dad. It's difficult and requires focus and deliberate life organization, but it's possible, and in fact quite rewarding and nice.

(Also, maybe the above makes it sound like I've had zero problems, but that's not true. It's been a long journey to figure out my priorities, with many mistakes and setbacks to being where I want to be. I have had times when I felt mentally less sharp, and worked to recover from those.)

Author here. I'm delighted to see this is still interesting / useful to folks. In case anyone is curious, here's a short origin story for this tutorial.

I learned Lua for a game jam - during the course of the game jam itself. Like many, I found it a bit weird at first. 1-indexing feels wonky (you get used to it, though); now that I use Python all day, I would miss syntax shortcuts like list comprehensions going back to Lua.

When I was a kid, I learned C from a little pocket reference book while building little self-motivated projects. I still think this is the best way to learn - have a project in mind, and just enough fundamental examples to get you there.

I threw together the Lua tutorial mostly in one sitting, spending some time to edit, and going back and forth on how to keep the whole thing runnable while still explaining how external modules worked. Shortly after publication, Adam Bard contacted me with his idea for learnxinyminutes.com. I said great idea and ok'd the inclusion of the tutorial on Adam's site. Adam invited others to write their own similar-style tutorials, and his site grew up quickly.

Lua itself is a truly beautiful language, though I understand this is not a popular opinion. It's beautiful because you can learn the entire language, through and through, quickly. Memory and comprehension of the language itself get out of the way, so the only hard problems left are your decisions about what to do with the language. (You ask "How will I?" vs "How can I?") Instead of being feature rich through added features, it's feature rich through good language design. I expect some will disagree with me, but as one comparison point, consider the complete syntax of Lua:

https://www.lua.org/manual/5.1/manual.html#8

versus, say, Chapters 11-14 of the ECMAScript spec:

https://tc39.es/ecma262/#sec-ecmascript-language-lexical-gra...

It's not apples-to-apples - the ECMAScript spec is spelling things out in detail, while the Lua section is an overview - but it's still a thought-provoking contrast.

I few years ago I put together a basic Mandelbrot terminal animation:

https://gist.github.com/tylerneylon/e70a684b88809f74168f3a36...

It runs nicely in Python 2 (and slower in Python 3).

Here's a nice code golf Mandelbrot that also uses no image library (it produces a bmp file): https://codegolf.stackexchange.com/a/23465

I was a very early stackoverflow user. I was excited and loved the idea. I still think it's a great idea, but I've come to abhor the police state moderation culture. To directly answer your question, yes, I completely agree that something has gone horribly wrong. I've forgotten more incidents than I can remember, but a few points I do remember: many good questions are closed; I've had questions hidden by people who (by their own admission) did not understand the tech being asked about but decided incorrectly the question was a duplicate; and I've seen some harmful cultural values widely supported on the meta pages.

My general impression is that some moderators seem to think of themselves as the gatekeepers to a promised land of knowledge, and the unwashed masses are just trying to ruin their pristine gardens, and must be aggressively pushed back. The truth is that stackoverflow can only exist because people like both receiving and giving help with their expertise. The site would be much better if it put this benevolence at its core rather than the guarded-perfect-garden model.

Yep! Completely agree that spaCy's docs are great, and have benefited from solid attention.

Our motivation in making this is to build up the site as a destination where you can map a conceptual ML problem to copy-and-pasteable code that immediately works and is readable / learn-fromable.

For the two of us (Mike Sall and I) creating the site, it's basically something we've often wanted in our day-to-day machine learning and data science work.

Thanks! Clearly we are fans of spaCy. I appreciate the offer on the pull request and would like to do so. Also, thanks for mentioning the clarification about the is_x attributes - I'll update the pages to improve the accuracy/clarity on that!

Cool connections. Do you happen to have a link for the digital art project?

Yes - what a great question to explore! Nice article, thanks.

I didn't know about that before. Looks very interesting (looking at wikipedia) — thanks for mentioning it!

Thanks! I'm considering writing a follow-up note on the process that went into the typesetting. The original source is a markdown-with-latex format that a tool called pandoc can work with. The style is primarily from tufte.css. The image work is mostly custom, but there are some re-usable approaches. If you're curious:

https://github.com/tylerneylon/math/blob/master/self_repl_fn...

I think you're right. Your version of the definition is more elegantly stated, and I may want to use it going forward, as long as a, b, and c still make intuitive sense (thinking about that part).

That's all the solid work I have so far. Tons of ideas for what could be next. For example, consider convolutions on sets generated by L-systems (eg the Cantor set). I suspect those might offer continuous, finitely-supported, linearly-shifted self-replicating fns (the current article is mainly about non-linear shifts; t_1 is only piecewise linear). Maybe L-systems even completely characterize that category of self-replicating fns.

Other ideas: * Higher dimensions (eg f1 + f2 + f3 = g with domain R^2) * Shifted fns f1, f2 whose sum has surprisingly small diff from their sum: ie small |C(f1+f2)-shift(f1)|; the normal fn appears to fall into this category

Thanks! The original source is custom js rendering to svg. The html version of the article contains png files based on that output; the pdf version contains vectorized versions. I got some solid design input from Mike Sall on the images for this article.

Edit: example: https://github.com/tylerneylon/math/blob/master/self_repl_fn...

This is independent work. Probably fits in with fractals and L-systems.

You're right that the g1 and g2 in section 2 are also scaled in addition to being shifts of f. I plan to clarify this in the text.

This work is inspired by the self-similarity of fractals, which is in some cases primarily an intuitive or informal idea: for example, the Mandelbrot set is considered self-similar, but I'm not aware (despite searching) of any simple automorphisms of the Mandelbrot.

I re-wrote ed in C once for fun:

https://github.com/tylerneylon/ed2

It uses the regex and readline libraries, as well as some container data types (array/map) called cstructs. The main C files sum to around 1,000 lines. It was an interesting exercise in learning some of the pieces of building a text editor.

A guy I work with created something similar a couple years ago. He used a different optimization technique which may be interesting to folks enjoying this: http://whoateallthepy.blogspot.co.uk/2014/01/shards-building...

If you have an array in Lua, which is a table used as a contiguous list of items, this is how you get the number of items:

    num_items = #my_array

Your qualm, apparently, is that this # is not built for all use cases of tables.

In C++, what operator tells you how many instance variables a class has? There is none, because that wouldn't make sense in many use cases. Lua tables, offering prototype-based inheritance, experience a similar conceptual ambiguity in the case where you want to take the length of an object.

Lua has made a design decision that the built-in length operator only works on arrays = tables-used-as-a-list and on strings -- cases without ambiguity. If you want the length operator to make sense for a hash map usage, it's a few extra lines of work for you. It's consistent with the langauge being small and flexible, a theme which Lua embraces with elegance.

It sounds like you're arguing that people wouldn't understand new results if they were more open. Most people I know well are educated/intelligent and could understand leading-edge research results within their own field, so I disagree. (And I don't work at a university!)

Even if only a minority of the public understood publicly-available research, having all results open could change the way the world works. It was hard to see how the information-sharing of the internet would change the world, so I can understand how this is hard to see as well.

You also hint that science deserves more wide-spread respect, and that the world be better off if people were better educated and tended to be more scientifically-minded. I agree with that!

Many mathematicians seem to have an attitude that a well peer-reviewed proof is simply correct and etched in stone for good. This seems to me like publishing programs - some of them thousands of lines long - and expecting them to be essentially bug-free without ever running them.

Von Neumann is not exactly making the same point, but he is arguing against the sense that math is a kind of purely abstract discovery that has escaped from the messiness of day-to-day reality.

One could ask: will we ever get to a point when all mathematical proofs are rigorously checked, so that we know they're correct? I hope we do get to a point when program-checked proofs are standard, but even then there's the possibility that a proof-checker itself is buggy. So, philosophically, it seems we may never justifiably feel certain beyond all doubt that any given proof is correct.

But we can have greater confidence in a proof verified by both expert humans and by a trusted program than in a proof simply reviewed by humans. So the effort is meaningful and worthwhile - all in the context of seeing math as a field as fallible as any other.

Felix Hausdorff committed suicide in 1942, after having put in quite some effort to escape Nazi Germany.

This was enjoyable.

I wonder if software exists that is aware of the overtones of each particular note of a given instrument. Theoretically, this would be a more precise way to pull out a chord from a recording without having to manually compensate for them later. Standard Fourier analysis is based on sine waves which aren't the best model for complex tones like that of a piano or a 12-string guitar.

Mathematically, the best approach I'm aware of in the long run would be to study the harmonic spectrum of the exact instrument as closely as possible, and then perform inner products against the spectrum of the mystery sound being analyzed. There's no guarantee that the note-vectors would be orthogonal (probably not), so the problem becomes more interesting in that you might want to perform a sparsity maximization on the resulting output - that is, look for the least number of notes capable of matching the recording within epsilon error in sample (or spectrum) space. Sparsity maximization is NP-complete, so a good way to approximate this is to actually perform L1-minimization using linear programming. Another approach might be to look for a balance between the absolute error term and the sparsity term.

I used to work at Medium on the product science team before this tool was created. I didn't work on it myself, although I can appreciate how well it works in practice. The simplicity-of-use of this tool, both as a data provider and as a visualization consumer, is incredible.

I can imagine some HN readers thinking they want more detailed graph control, but the few options + the defaults on this tool cover a surprising range of use cases. I think you have to try it on your own data to appreciate the subtle decisions that went into the design.

This interface is solid. So many easy-to-use graphing tools have poor defaults. This is perfect for the use case of a quick visualization for queried data or (even better) live data you'd like to dashboard.

The Fourier transform is the map between sound waves and music as written on a staff.

That's an informal statement with some hand-waving since, for example, music notation is discrete and sound waves aren't, but that's the main idea.

Thank you!

The thought experiment with infinite hats is interesting.

It sounds funny to ever say "the AoC is wrong," though, because axioms are things you get to choose as true or false. They are essentially subjective, and in practice are as "socially true" as they are commonly used. I guess the author is arguing that the more intuitively satisfying set of axioms excludes AoC.

I personally disagree, although I admit there's still room for debate because of compelling arguments on either side (most of which are appeals to intuition by the nature of the problem).

Here's a small counter-argument: Since we're appealing to intuition, consider that we're asking a countably infinite number of people to (a) see and reason based on seeing an infinite sequence in front of them; and (b) all memorize a specific function which is essentially from all the reals in [0,1] to a subset of [0,1], which we may think of as binary numbers with infinite expansions. (The intuitive point is clear without having to worry about 0.01111... being the same value as 0.1). This function is insanely weird and hard to describe succinctly; basically, it must be memorized. My intuition tells me that as soon as we expect agents to perform these feats, they are no longer humans and no longer Turing machines. Since this memorized function is so weird, they are even beyond some kind of infinite-time Turing machines with finite memories (or infinite-memory TMs with finite time). I think they are even beyond infinite-time Turing machines with countably-infinite memories, although I haven't proven this carefully.

In the long run, I suspect an alternative axiomitization (or possibly an elegant classification of set theory statements) will shed light on these problems, somewhat analogous to the way we now consider non-Euclidean geometry as interesting and "true" in its own universe that we mentally separate from the Euclidean plane.

[ps Not all comments about axioms are appeals to intuition. E.g., you can prove some are independent or dependent. But arguments along the lines of this post are appeals to intuition when used for/against AoC.]

I'm a big fan of self-analytics. I also like small-but-useful command-line tools of this nature. This is pleasantly documented and purpose-driven.

If you're looking for something more focused on short daily summaries of what you've done, I wrote a similar tool called wj:

http://tylerneylon.com/a/wj/

It's more about organization and presentation than privacy. For example, it can generate a tex file that converts your entries into a nicely formatted pdf file.

I also like the automated self-analytics Stephen Wolfram uses:

http://blog.stephenwolfram.com/2012/03/the-personal-analytic...