HNHacker News
TopNewBestAskShowJobs

tylerneylon

2,127 karma · joined July 3, 2011

code: https://github.com/tylerneylon personal: http://tylerneylon.com

Email: tylerneylon@gmail.com

submissionscomments
tylerneylon··on Elsevier Publishing Boycott Gathers Steam Among Academics
Two thoughts on 'this does not go far enough':

(1) It is a step in the right direction, and it's a step that is small enough for more people to take. Why does thecostofknowledge.com have over 2000 names after a week, while research without walls has under 500 after several months?

http://www.researchwithoutwalls.org/

I think the difference is that thecost is a focused effort that asks less from researchers. Perhaps this would / will be easier when there are more well-regarded journals using an acceptable publishing standard.

(2) I had an interesting discussion with rms about the core problem with for-profit academic publishers. He made some excellent points arguing in favor of what he called "redistributable publishing" (which we could also call "free-to-copy publishing" ?) The argument is that some definitions of "open access" do not ensure the right of any individual to pass on articles, and, as we can see with software and other electronic data, this right can make a huge difference. So, from that perspective, some aspects of open access could be said to not go far enough.

I would love to see as much access and freedom as possible for any research produced by people who want it to be free. I think the best way to do this involves a bit of understanding and pacing for those whose careers depend on how they handle this move.

tylerneylon··on Fun with math: Dividing one by 998001 yields a surprising result
The general any-number-base-b rule here (as others have noticed in base 10^k) is that

1/(b-1)^2 = 0.0123456... (where '1', '2'.. are base b digits).

The original post is this fact in base 1000.

Proof for any base: 1/(b-1) = 1/b + 1/b^2 + 1/b^3 + .. = 0.11111.. (base b).

So 1/(b-1)^2 = 0.11111.. * (1/b + 1/b^2 + 1/b^3 + ...) = 0.012345... QED.

Richard Feynman beat us all to the punch here by noticing that 1/243 = 0.004115226337..., a fact which he wrote in a letter from a secret lab to someone in the outside world, and which put him under suspicion of sending secret messages! That gem of a fraction turns out to be a result of the above stuff as 1/243 = 111 * (1/999^2) + 4/999.

Here's a slightly more detailed explanation:

http://tylerneylon.com/b/archives/51

tylerneylon··on The Cost of Knowledge: Researchers taking a stand against Elsevier.
Agreed. I just added email verification. I plan to check that the emails match the names and school while it's practical to do so (while the list is not crazy long).
tylerneylon··on The Cost of Knowledge: Researchers taking a stand against Elsevier.
I think Elsevier is a good position to act as a kingpin in the math world when it comes to the academic publishing racket. I'm not sure which publisher is worse than others, but I do think it's more likely that success will come against a particular target (more researchers will take the pledge), and that things would change with all publishers as a result. Elsevier has a long list of well-respected math journals, so a victory with them could be most meaningful in the eyes of the mathematicians who choose where to submit their papers.
tylerneylon··on The Cost of Knowledge: Researchers taking a stand against Elsevier.
Agreed, it's in the works. I'm balancing this with my startup.
tylerneylon··on The Cost of Knowledge: Researchers taking a stand against Elsevier.
I created this page this weekend as one small thing I could do to help the math world and academia in general with this problem.

Would anyone be willing to help me add improvements to the page? It's become popular more quickly than I expected, and I would appreciate the help. I've temporarily added my email address to my hacker news profile, so anyone can contact me. Thank you!

tylerneylon··on TVs are all awful
I disagree. 400ms is noticeable lag to anyone, still so against 100ms. Run this in bash, and imagine this lag in a game you're playing, as the time from when you hit a button on a controller to when you see the response on screen:

$ sleep .1

$ sleep .4

tylerneylon··on K-Means Clustering and Art
I wrote some similar code in python recently -- I think color visualizations like this are both fun and potentially useful for certain image manipulations.

This includes a readable 36-line implementation of k-means clustering that could be shorter if one wanted to play some code golf :) I used a pie chart layout, with pie slices proportional to their corresponding cluster sizes.

Code: https://github.com/tylerneylon/imghist/blob/master/imghist.p...

Sample images: http://blog.zillabyte.com/post/11193458776/color-as-data http://blog.zillabyte.com/post/13141231882/hue-histograms

If anyone else is interested in this stuff, Austin A made a great suggestion on the original post to use the Lab colorspace.

tylerneylon··on Scientists' proposed calendar synchronizes dates with days
The 7date doesn't say much about how we count years. It's all about how you count days and other sub-year units. You could combine it with any method of counting years that you like. I defaulted to using the year system I'm used to because it's convenient.
tylerneylon··on Scientists' proposed calendar synchronizes dates with days
This problem has been seen before, when countries adopted the Gregorian calendar at different times:

http://en.wikipedia.org/wiki/Dual_dating

tylerneylon··on Scientists' proposed calendar synchronizes dates with days
I've been using an independent calendar for my own records for over a decade now - it's the day-of-year, written base 7. It fits in with weeks; if we reset weeks every year (why not? we do worse things already with leap days and seconds) then dates become equivalent with weekdays. One number expresses the day-of-year, day-of-week, week-of-year, and month-of-year, using a 49-day month.

http://tylerneylon.com/b/archives/122

I don't really expect many other people to use this, despite being a better designed system. I think hardcore math/cs people would like it, though. I've written some small pieces of code that work with this system, if anyone's interested.

tylerneylon··on Timsort
A bad implementation of quicksort has O(n^2) worst case, and that's not really a fair comparison. I'm not defending quicksort, just pointing out that a fair comparison should keep in mind that we can easily select better pivots to achieve best-, average-, and worst-case complexity of O(n log n) for quicksort.

Comparing sort algorithms like this shows that we really care about more than just time complexity. We care about memory used or swapped down to the byte, quirks on popular processors, and expected run times on real-world data.

tylerneylon··on DressRush Pitch Deck
How were these slides created? Basically I'd love to be able to make a similar pitch deck for my startup.

A little embarrassed to ask b/c I'm guessing it may be obvious to those who use these tools every day (not me, though).

tylerneylon··on DressRush Pitch Deck
Some people are suggesting this number is too high. If anything, it seems low to me - at least based on my personal experience. Imagine throwing a big, elaborate party. How much would that cost you? Maybe $3k, including renting a space, maybe a DJ, a bartender, etc? Everything for a wedding is 10x what it should be. So if a reasonable price is $3k, you're paying $30k.

This is without the price of the ring and honeymoon.

My source is personal experience, and that of my friends, but also see: http://www.costofwedding.com/

tylerneylon··on For HN: Open access to scientific papers, a summary of the state of play
How easy would it be for an open access startup to be profitable?

I'd like to see the cause supported and this sounds like a fun way -- but it's not clear to me how that would work as a business.

tylerneylon··on Why code review beats testing: evidence from decades of programming research
Most of this content is already in Code Complete.

There are two anti-bug ideas I strongly agree with: bugs are found by use of the code, whether through testing or actual use; and that great unhindered programmers tend to write solid code. It's very logical that bugs will be found by usage - this is basically the test-driven reasoning. As far as programmer quality, it is also logical, but it is underemphasized in the article.

I distrust articles that emphasize one-size processes for software development. In the end, there's talent and something like culture. Government agencies that can't go out of business follow policy, and provide results just above the legal required standards. Old companies with seniority-based hierarchies feel similar to code monkeys. A small company with respectful, productive, and creative builders is ripe for a culture of sincerely-desired quality. This is what I mean by the importance of culture over process.

tylerneylon··on Peano arithmetic is probably inconsistent
If this is true, it would be a bigger deal to logicians than a solution to P vs NP.

My guess is that there will end up being a mistake somewhere (based purely on statistics of huge claims without full posted proofs), but it seems too early to say anything definitive. Terence Tao has guessed it is wrong, which is evidence against it, but then Nelson acknowledged Tao's remarks and said it isn't a problem, which is interesting. This is making me wish I was more of an expert so I could decide for myself.

tylerneylon··on Ask HN: Is this data startup idea viable?
Yes, I think it is, although I see a few challenges.

It is a huge amount of work to integrate disparate data sets and build a system that can efficiently handle many customers' requests.

It may be difficult to show customers how they can actually use it. Comparing two stocks isn't cool. You know what's cool? Comparing all stocks with realtime updates.

For the second problem, I mean that users won't use their imagination much to figure out what your product does. You have to spell it out in concrete, actually useful examples. This is a challenge between engineer-thinking, where general functionality is king, and user-thinking, where one particular goal and the path of least thinking/work is king.

tylerneylon··on PROGRAMS WITH COMMON SENSE (J. McCarthy, 1959, pdf)
This paper is historically very interesting.

However, I think (and I'm not alone) that it set AI research off in a bad direction. It's not that formal logic manipulation toward common sense is not important, but rather that it is not the skeleton key to the artificial intelligence dream. It's hard for me to _prove_ that statement, but we can see the palpable disillusionment of researchers who've tread this path.

If I were personally doing AI research now, I would begin with emotions. Humans act because they want something, and they only use logic as a means to an end -- and certainly not perfectly at that. If our goal is to achieve a thinking program, a rational approach would be to walk down the course evolution has found -- simulate the mind of a fly, a toad, a dog, and a monkey first. It may be tempting to design something top-down, but that presumes we understand the top very well, and I don't think we do. (By "top-down" I mean from behavior to code; I'm sidestepping the biochemistry of neural activity, whose study implies we need to know the brain's hardware before we can emulate its software.)

As a side note, some folks use AI and ML (machine learning) interchangeably, but there are really two goals around. I see McCarthy as wanting to create a program that could pass the Turing test (AI), while support vector machines and other ML tools are designed to perform advanced statistical analysis of data, along the lines of extrapolating a pattern. Related but different goals, and this paper has been viewed historically as about AI.

tylerneylon··on How to Make a Clock Run for 10,000 Years
I love the dream of building things to last. If humans left earth this instant, what would remain useful in 1000 years? Most of what we use today would perish, so there is a sense of overwhelming disposability or high-maintenance around our lifestyles.

Rather than viewing this disposability as a weakness, perhaps we can see it as an element of wise design. By analogy, the human body leans heavily towards self-repair rather than monolithic permanence. Perhaps it is a good thing that our day-to-day designs depend so much on active use and maintenance. If no one continues to use it, why does it need to keep working?

tylerneylon··on How did academic publishers acquire these feudal powers?
I do, and they require $500/year for library use by non-students/staff.

Even if some people are lucky enough to have university library access, the problems caused by a broken publishing system still pose a barrier to many independent people.

tylerneylon··on How to Design Classes (in Ruby, Python, Java or any OOP)
This book is too long for the subject it purports to impart.

Good coding is not a craft of many secrets; rather it is an art of a few core principles. Mastering them is a matter of direct experience. I believe a good book can point a programmer in the right general direction, but it is up to her to go the distance.

tylerneylon··on [dead]
I bought a desk-sized Lack table from Ikea, and took it to Home Depot to saw the legs down, so that it added around 12" to my standard desk. Now I can stand while working. Assuming you have an external monitor / keyboard / mouse, the only thing left is to put your monitor on some books to get it closer to eye level. I find this is ergonomically very comfortable if the desk & monitor are both at optimal heights.

Total cost: $40.

tylerneylon··on A History of Bayes' Theorem
It has always bothered me that Baye's theorem is a named result, instead of just "the law of conditional probability," say. As soon as you formally define conditional probability, it is obvious.

To be fair, though, I think he discovered it before probability theory was formalized, and I can see how it may have been non-obvious at the time. Sometimes the best progress makes challenging insights appear obvious in hindsight - nice work, Laplace!

tylerneylon··on How did academic publishers acquire these feudal powers?
The obvious negative this has on folks outside of research-level academia is a significant contribution to tuition prices, which seem to be rising at about 5% per year.

But a less obvious - and personally very painful - consequence of greedy publishers is the inability to do serious academic research outside of academia/industry. I have the ability (math PhD) and the will (have published a few papers post grad school, though it's hard to find time), but I have all but given up due to lacking access to books and articles behind these paywalls. Yes, you can find a decent chunk of articles online -- but very often there are one or two (or more!) key papers you _need_ to read to be at the front of a field, and one of those will be behind a paywall. The worst part is that I never know how truly useful an article will be before reading it, so in the few cases where I've payed I find that only a small percentage of the time was it worthwhile.

In short, this system essentially kills research outside of academia / industry.

tylerneylon··on Steve Jobs Resigns as Apple CEO (Official Letter)
Whatever happens next, this is the end of a great era for Apple.
tylerneylon··on BSON and Data Interchange
See also bert-rpc, which is a cross-language/platform serialization standard along with fast rpc support. This is what github uses:

http://bert-rpc.org/

tylerneylon··on Apple's iOS unaffected by malware; Android exploits surge by 76%
I like the installation options on a mac: you have the official app store, but you don't have to use it.

In theory, Android could also offer different marketplaces, one where the apps are carefully reviewed, and another where anything goes. Keeps the choices, but also supports safe downloading to those who want to be careful. (Ok, Android already _does_ have multiple app stores, but I don't think of any particular one as "the official reviewed marketplace," which is what I'm after.)

tylerneylon··on Hey, Bethesda Let’s settle this
This is a great solution because it can settle things quickly with less stress and let everyone get back to work.

Gamers are _not_ going to confuse Notch's and Bethesda's use of the word "scrolls," so I don't see why either company should take this too seriously. (Re-naming a game is less work than a drawn-out lawsuit.)

tylerneylon··on Parse (YC S11) adds User Accounts to Android.
It would be cool if there were custom support for linking accounts, such as with facebook. That code is never fun to write, but common.
← PreviousPage 6 of 7Next →