# niklas@web: /var/log/

Trying to escape local optima on a random walk through life.
• #1622620264 2021 Jun 02, 09:51 – Our genes take the greedy local maximum.

• #1621845157 2021 May 24, 10:32 – Sanger sequencing

This is pretty cool.

• #1621452087 2021 May 19, 21:21 – Classical Mechanics Lecture

Recently, I’ve started reading the book The Theoretical Minimum after noticing my huge physics knowledge gap while flipping through Thinking Physics. However, to me the lectures seem to be rather mandatory than supplementary, because the book alone won’t save me from drowning in confusion about e.g. the introduction of the Lagrange formalism, which is done basically in passing. I really like the lectures though! Both the book and lectures are also pretty mathematical, so that’s a plus.

• #1619074721 2021 Apr 22, 08:58 – Proselint

Proselint is – as the name might suggest – a linter for prose, that may help you write better texts. Install it from Github or visit their website.

• #1619074461 2021 Apr 22, 08:54 – That Alien Message

A parable on AI Alignment that helps grasp how big the difference between human intelligence and AGI can be.

• #1618488571 2021 Apr 15, 14:09 – Detexify

In case you’re looking for a TeX symbol and don’t know what to search for, try drawing it!

• #1618429351 2021 Apr 14, 21:42 – Fermi Estimates

Up until now, I’ve only done a few scattered Fermi estimates here and there, but they’ve always been pretty interesting and fun to do. This might actually make a handy pastime for waiting times, longer travel times or hikes with friends. I just stumbled upon a cool post about it.

• #1618392317 2021 Apr 14, 11:25 – Predictive Coding Approximates Backprop along Arbitrary Computation Graphs

As already approached in my most recent post, this paper provides further evidence that the workings of connectionist models and human brains might actually be very similar (modulo implementation details). It demonstrates that a possible method of operation of human brains – namely predictive coding – actually converges to backpropagation, the standard way artificial neural networks do their learning.

Predictive coding is an influential theory of cortical function in theoretical and computational neuroscience. […] [The presented results] “demonstrate that predictive coding converges asymptotically (and in practice rapidly) to exact backprop gradients on arbitrary computation graphs using only local learning rules.”

• #1618207380 2021 Apr 12, 08:03 – An Untrollable Mathematician

On logical uncertainty, the interplay of logic and probability theory, Gödel and trolling – what more could one want?

Also, could we use such vivid illustrations on a regular basis, please?

• #1617438226 2021 Apr 03, 10:23 – Pascal’s mugging

“Give me five dollars, or I’ll use my magic powers from outside the Matrix to run a Turing machine that simulates and kills $$3 \uparrow \uparrow \uparrow \uparrow 3$$ people”, Wikipedia.

• #1616661297 2021 Mar 25, 09:34 – An Intuitive Explanation of Bayes’s Theorem

P(A|X) = [P(X|A) × P(A)] / [P(X|A) × P(A) + P(X|¬A) × P(¬A) ]

• #1615754798 2021 Mar 14, 21:46 – Alternatives to backpropagation

• #1614798026 2021 Mar 03, 20:00 – Only Metadata

As NSA General Counsel Stewart Baker has said, “metadata absolutely tells you everything about somebody’s life. If you have enough metadata, you don’t really need content.” When I quoted Baker at a recent debate at Johns Hopkins University, my opponent, General Michael Hayden, former director of the NSA and the CIA, called Baker’s comment “absolutely correct,” and raised him one, asserting, “We kill people based on metadata.”

• #1614416865 2021 Feb 27, 10:07 – Signs & Symbols

This might be – alongside the Luzhin Defense – one of my favorite pieces by Nabokov. It’s really short, you can read it here.

• #1612266327 2021 Feb 02, 12:45 – Metaculus

You might be wondering why it’s gotten so quiet on here.

Metaculus is a community dedicated to generating accurate predictions about future real-world events by aggregating the collective wisdom, insight, and intelligence of its participants.

Some months ago a friend introduced me to Metaculus, and yesterday Scott Alexander also wrote some paragraphs about it, so I thought I’d share it here as well.

• #1612207629 2021 Feb 01, 20:27 – “Rationality: From AI to Zombies” Podcast

• #1612179380 2021 Feb 01, 12:36 – Google Scholar Browser Extension

Great way to quickly pull up pdfs from Google Scholar while surfing. Link.

• #1612130179 2021 Jan 31, 22:56 – Who wants to do some p-value hacking?

Or maybe just lose their trust in Scientific journals?

Jokes aside, of course there’s more to it than that, and I really enjoyed this article about some widespread problems in science right now.

• #1611830900 2021 Jan 28, 11:48 – Spaced Repetition done right

While reading through this excellent article on self-teaching, I noticed that a lot of my current endeavors could be greatly simplified by actually knowing my facts: Spanish grammar rules, some connectors for the TOEFL test (yes, it does seem to be a redundant acronym), the definitions for my upcoming Security exam, the stuff I read for fun in various textbooks etc. etc.

Thus, I’ve decided to give Anki another go, and do it right this time.
Up until now, I’ve only ever used Anki while studying for exams, then stopped using it after I passed. This kind of never-ending spacing is – of course – not really the intended use of spacing in Spaced Repetition-techniques. So now, I’m trying to actually build a habit instead of just using Anki to cram in all the facts during the weeks before an exam.

Here are some practical tips for designing cards for spaced repetition.

• #1610661597 2021 Jan 14, 22:59 – Machine Learning: The Great Stagnation

Machine Learning PhD students are the new Investment Banking analysts, both seek optionality in their career choices but differ in superficial ways like preferring Meditation over Parties and Marijuana & Adderall over Alcohol and Cocaine. A Machine Learning PhD is now just an extended interview for FAANG.

Harsh but interesting article on the current state of Machine Learning research.

• #1609663512 2021 Jan 03, 09:45 – Experiment more

Except for very young children, people categorically Explore too little and stagnate in local optima.

• #1608386912 2020 Dec 19, 15:08 – An introduction to hidden Markov models

To further refresh my knowledge of HMMs, the main problems that have to be solved and the algorithms used for those, I’ve worked through this amazing paper (pdf). It’s a great (and very thorough) explanation, there’s only a typo in the recursive equation of the backward-algorithm on page 6:
Instead of $$\beta_{t}(i)=\sum_{j=1}^{N} a_{ij}b_{j}(O_{t+1}(j)$$, I think it should be $$\beta_{t}(i)=\sum_{j=1}^{N} a_{ij}b_{j}(O_{t+1})\beta_{t+1}(j)$$.

• #1608229406 2020 Dec 17, 19:23 – A friendly introduction to Bayes Theorem and Hidden Markov Models

I’m currently doing some basic research for my Bachelor’s thesis and I found this video tutorial pretty helpful as a very first, very basic introduction and recap: Link.

• #1607777826 2020 Dec 12, 13:57 – μCovid

Online calculator to get a better feeling for Covid related risks of certain activities. It’s pretty configurable and well backed by recent scientific results.

• #1607204554 2020 Dec 05, 22:42 – 8,760 Hours

In case you’re planning 2021 soon, take a look here.

• #1607016854 2020 Dec 03, 18:34 – Sometimes, code is documentation

I just love looking up things in Haskell’s Prelude. Quickly get the information you need for using it and instantly see how it’s implemented as a bonus.

• #1606814618 2020 Dec 01, 10:23 – Harry Potter and the Methods of Rationality

I don’t particularly like Harry Potter and I’ve never read a fan fiction before, but this is definitely not what I’d call the typical “Harry Potter fan fiction”. I can’t believe I’ve waited this long to try it out. If you’re interested in rationality and a funny and informative story, you can find it here. I highly recommend the audiobook version.

• #1606814417 2020 Dec 01, 10:20 – Advent of Code 2020

Starting today!

• #1606745078 2020 Nov 30, 15:04 – Justice Der – Covers

Some pretty relaxing guitar covers.

• #1606324520 2020 Nov 25, 18:15 – Hammertime!

Want to do yourself a favor and get better at living life?
Complete the Hammertime Sequence for thirty days of instrumental rationality practice.

• #1604845992 2020 Nov 08, 15:33 – How does consciousness even make sense?

I don’t think the current state of “artificial intelligence” really has proven that it earns the great title of intelligence, as I believe it’s all still just a sophisticated application of statistics on large amounts of data. I prefer the title “machine learning”, as in my opinion that describes the process of adjusting the parameters of statistical methods based on the given data adequately.

I’m not even sure if intelligence and consciousness can be simulated by a computer. Because if they could, then the speed of execution surely wouldn’t matter to that fact, right? And if speed didn’t matter, one could just as well represent the (deterministic!) calculations of a finite computer on a piece of paper or by arranging some stones on a large field. Granted, it’d be somewhat slower than a modern computer and the paper would have to be sufficiently large, but in the end flipping bits, drawing on paper, and moving rocks in a systematic way is just the same when it comes to representing computation. So that’d mean if we arranged a bunch of stones on a large field in a certain pattern and then used some fancy rules to move them around, we’d create consciousness?! I can’t really believe that’s true.

But how is a human brain any different?? In the end it’s also just biological wires exchanging electricity (+ some chemistry added to the process)…
I can’t really grasp that. Do my thoughts make sense? Where’s the flaw?

• #1604502578 2020 Nov 04, 16:09 – Long live jumplists

The jumplist (:help jumplist) has recently become one of my favorite features in not only vim itself, but also vim-based programs, especially my pdf-viewer zathura. After following intrinsic links in a pdf, I no longer have to search for the previous position, but can just press ^O and continue reading where I jumped from. The same of course goes for searching or any other jump performed.

• #1604317803 2020 Nov 02, 12:50 – Tensorflow Neural Network Playground

Interactive visualization of the training process of small NNs. Link.

• #1603235619 2020 Oct 21, 01:13 – Fail2ban & Free tips for intruders

While going through my log files, I just noticed there were automated attacks on this server all around the clock, so I decided to upgrade my security a bit.
I think installing and enabling Fail2ban is a good first measure and should stop the vanilla script kiddy from performing those attacks. I don’t think knowing about this new security measure will help anyone bypassing it without using disproportionately large guns (Note for intruders: you’d have to distribute your simpleminded brute-force attack – see I’m even cooperative!); they probably didn’t even take the time to read my free tips for intruders…
P.S.: As a result of trying out the new defense mechanism, I now have to wait a while before I can update this log. But it works!

• #1603197403 2020 Oct 20, 14:36 – Backstreet’s back & Game Theory

The beginning of this month was mostly spent learning about the wonderful world of Numerical Mathematics in order to pass one of my last exams in my bachelor degree course, so I wasn’t really exploring new stuff (except all kinds of matrix decompositions) to share on here…
I also felt a bit sick over the last few days (no Corona though, I guess), but I’m feeling a lot better today, so I’m back at doing some light studying. Today: Game theory! This paper seems like a pretty good introduction to me, although I was very confused at first by some of the notation used; some elements are given the same name as the sets they’re from?! One thing I found particularly intersting is that you can draw little arrows across each dimension of your normal form games’ tables and they always point into the direction of an equilibrium (Cycles mean there is none)!

• #1601548268 2020 Oct 01, 12:31 – Advanced Linear Algebra

Check out this YouTube channel about Advanced Linear Algebra, I’ve watched the series on Householder transformations and it was pretty useful.

• #1600867320 2020 Sep 23, 15:22 – The infamous ghetto indoor pool

Top-notch entertainment. No penguins were harmed in the making of this post.

• #1600791863 2020 Sep 22, 18:24 – Donsol – A Dungeon Crawler based on cards

Website and a video.
I’ve also written a short manual for it: Web version and PDF version.

• #1600778210 2020 Sep 22, 14:36 – Orca and more stuff by Hundret Rabbits

Orca is a pretty minimal, esoteric programming language for generating sounds.
Here’s a live coding tutorial that explains the basics. Here are some demos by Devine himself.

Wow, the more I see, the more I’m impressed by Hundred Rabbits’ lifestyle and projects.

• #1600381416 2020 Sep 18, 00:23 – North Pacific Logbook

Some logbook entries of two people sailing from Japan to Canada. Check out the rest of their website as well!

Fun fact: Today I’ve been to my first demonstration, it was pretty interesting. No burning cars though.

• #1600251276 2020 Sep 16, 12:14 – Boarding pass hacking

When you browse Instagram and find former Australian Prime Minister Tony Abbott’s passport number, the fun to read Do not get arrested challenge 2020 of Alex.

Also check out this post here: Hacking your neighbor’s Wi-Fi.

Oh, and while you’re at it, check out Stalking your Facebook friends on Tinder as well!

• #1600188375 2020 Sep 15, 18:46 – Comparative advantage and when to blow up your island

Funny and interesting article.

• #1600075551 2020 Sep 14, 11:25 – Less wrong & Learn anything

In the last days I found two great resources for learning awesome stuff:
LessWrong is a community with the goal of developing, training and applying rationality to real-world problems. Learn anything is a search engine combined with a knowledge graph that serves you study guides to many different topics.

• #1599818957 2020 Sep 11, 12:09 – There is no ignorabimus

Wir dürfen nicht denen glauben, die heute mit philosophischer Miene und überlegenem Tone den Kulturuntergang prophezeien und sich in dem Ignorabimus gefallen. Für uns gibt es kein Ignorabimus, und meiner Meinung nach auch für die Naturwissenschaft überhaupt nicht. Statt des törichten Ignorabimus heiße im Gegenteil unsere Losung: Wir müssen wissen — wir werden wissen!

We must not believe those, who today, with philosophical bearing and deliberative tone, prophesy the fall of culture and accept the ignorabimus. For us there is no ignorabimus, and in my opinion none whatever in natural science. In opposition to the foolish ignorabimus our slogan shall be: We must know — we will know!

– David Hilberts address to the Society of German Scientists and Physicians, in Königsberg (8 September 1930).

There is a very interesting recording of this address and a great book on Hilbert.

• #1599235651 2020 Sep 04, 18:07 – Write Yourself a Scheme in 48 Hours

I have to recommend this great tutorial on implementing Scheme in Haskell. It’s interesting, straight forward, highly informative (you’ll learn about Scheme, Haskell and interpreting) and also very rewarding.

• #1599163524 2020 Sep 03, 22:05 – The Future of Neuralink

Lex Fridman (interesting guy by the way) shares some exciting thoughts on the future of Neuralink (and eventual competitors). Watch on YouTube.

• #1599076365 2020 Sep 02, 21:52 – Discipline in Thought

• #1598824508 2020 Aug 30, 23:55 – Neuralink/Memex

Here’s a short summary of Elon Musk’s Neuralink presentation on YouTube and here’s a link to a video presenting a pretty cool Memex project, in case you’re not quite ready for the real deal™.

• #1598820150 2020 Aug 30, 22:42 – The Recursive Universe

I just found this cool GIF of a recursive version of Conways Game of Life (link).

• #1598736317 2020 Aug 29, 23:25 – Judgify Launch

So today I spent a few hours creating a web app that shows you how basic your taste in music is: Get judged on Judgify.
I also spent some time bricking the firmware of my mechanical keyboard. :(

• #1598556576 2020 Aug 27, 21:29 – A great overview of attention mechanisms and models

In this article, Lilian does a great job at explaining this (pretty fascinating) topic.

• #1598454610 2020 Aug 26, 17:10 – Introduction to Convolutional Neural Networks

Stanford offers a great overview on ConvNets on their course website.

• #1598444048 2020 Aug 26, 14:14 – Patents considered harmful.

I just found out that not only does Google hold the patent on dropout, but also on processing images using deep neural networks and Microsoft even holds the patent to train ConvNets on GPUs. What is wrong with the patent office?!

• #1598393069 2020 Aug 26, 00:04 – Toccata

Quite interesting visualization.

• #1598186260 2020 Aug 23, 14:37 – Understanding LSTM Networks

Check out this short article, in which Christopher Olah explains LSTM networks and shows some helpful images.

• #1598024665 2020 Aug 21, 17:44 – An Interactive Introduction to Fourier Transforms

Cool website to get a basic understanding of Fourier transforms.

• #1597702710 2020 Aug 18, 00:18 – Tracking computer usage from the command line

Thyme provides a pretty neat CLI to track your active/visible windows. I’ll do this for at least some days to get a better perspective on how I spend my computer time (by the way, I also recommend to track your nutrition for one to two weeks, just to get a general estimation and feeling for your diet). There was a bug however; the program didn’t terminate the current time slot leading to huge slots of tracked time while the computer was actually in standby or shut down, which I fixed in my fork. I also did a pull request.

• #1597320581 2020 Aug 13, 14:09 – Online tool to export epubs from Wikisource

Wikisource is a repository for texts in the public domain. I just found this tool to export the texts as ebooks.

• #1597048233 2020 Aug 10, 10:30 – Some thoughts on the computability of numbers and the continuum hypothesis

So last night I wondered if all real numbers were computable.
Because if that was the case, then every real number could be matched to a turing machine (the one that computes it), which in turn could be matched to a natural number (its Gödel numbering). But this would imply $$\aleph_{0} = \aleph_{1}$$, which is a contradiction, because the reals are isomorphic to the powerset of the natural numbers. So there are reals that are uncomputable.

Is this the case because computations have to be deterministic and finite? But you can calculate π? Or are approximations invalid? Because calculating π exactly is not finite, is it?

Okay, so I took a look and I’m really not surprised that they just brought up the Halting Problem again.

• #1597009532 2020 Aug 09, 23:45 – New stuff!

I’ve just uploaded a python script supporting the theory of my latest article on investing with some statistical methods and the implementation of the presented strategies. I’ve also added my guitar recording to the Goals 2020 article, you can listen to it here.

• #1596986971 2020 Aug 09, 17:29 – A short and good recap of Backpropagation

Presented in a 10 minute video by 3Blue1Brown on YouTube.

• #1596709260 2020 Aug 06, 12:21 – 5D Chess With Multiverse Time Travel

It’s kind of hard to understand by simply watching the trailer on their website, but this video helps a bit with understanding the new concepts. Also a nice idea by zw123456 in the HN comment section:

I always thought a cool chess variant would be quantum chess where you can move a piece to more than one square and assign a probability to each position. Your opponent could do the same. But I never worked out how it could work from there.

• #1596645228 2020 Aug 05, 18:33 – Some Fundamental Theorems in Mathematics

An expository hitchhikers guide to some theorems in mathematics. PDF.

• #1596550779 2020 Aug 04, 16:19 – CLI tool to directly convert LaTeX to png images

Clone it from Github.

• #1596447502 2020 Aug 03, 11:38 – Insights into ransom negotiations

Indeed, very rare and interesting insights into how these deals take place. Although I can’t believe each one is handled so professionally by both sides. Twitter thread.

• #1596445969 2020 Aug 03, 11:12 – Comments on A philosophical difference between Haskell and Lisp

I just stumbled over a discussion on HN and found two special comments I wanted to share:

There is a strange… kind of poetry with Haskell. It is like math on wheels, math applied to procedures, math with… time. Its appeal to me is like the appeal of math to me, not like real analysis math but abstract algebra math. The beauty, the purity of mathematics of younger days that once became lost after encountering the sad complexities of the world. Understanding every little aspect and being able to prove every part is now a luxury, and interfacing without real understanding is the more practical approach in the turbulent waters of poorly connected technological and social systems.

But still there is hope and there are dreams. We like to drench ourselves in dream qualia sometimes, and Haskell and pure math are that medium. The abstractions of it, the consistency of it, the purity of it… When Haskell is called a pure language, it almost goes beyond the static definition of functions being pure, and describes the general feeling that occurs when writing Haskell. You feel pure. You feel like you are taking these small parts and creating greater parts in an elegant buildup of abstractions, traversing one level higher and one level lower at your whim.

Lisp… maybe it’s the parentheses, maybe it’s something else… it never really caught onto me like Haskell did. Haskell feels pure and dream-like and perhaps unsuited to the world where (if you really get down to it) abstractions and types are just useful ‘human’ inventions and unfit for every usage. The world is for getting down and dirty, and mathematics, or at least the pure side of it, really isn’t. The representative mathematics of Haskell is Category Theory, and it is just as far from the level of “real” as it can be. More abstract than abstract algebra, if you will.

Abstraction itself is an intellectual operation that is also rooted in emotional detachment. Perhaps Haskell represents that kind of ideal in a modern world where practicality pays before purity.

– avindroth

This is a matter of feeling, completely irrational, but just my experience.

When I program in lisp I get the same feeling when I’m solving an ODE by hand, or a Diophantine equation, or designing a numerical method to approximate the solution of a PDE, or finding the Euler-Lagrange equations of a physical problem. The thing is real and it gets shit done really fast; it is just exhilarating. Moreover, lisp macros are so dirty and fun that using them feels like kinky sex.

Haskell is like category theory. Sure, it is pure and general, and you can create set theory and the rest of math from it. But it is certainly the most boring thing that I can think about.

But hey, whatever floats your boat.

– enriquto

• #1596103561 2020 Jul 30, 12:06 – Talking in front of people

Yesterday was the first time I talked in front of 100 people and had to use a microphone. It was an exciting new experience, and I actually wasn’t as nervous as I thought I’d be. I’m looking forward to the next opportunity, it was a lot of fun!

• #1595678083 2020 Jul 25, 13:54 – Sandy Maguire: Expert Level Vim

A nice little talk by Sandy about the basic concepts of Vim and the language that is used for describing the process of editing text. YouTube link. The title is a bit misleading though.

• #1595445597 2020 Jul 22, 21:19 – A quote I found in the SICP book

I think that it’s extraordinarily important that we in computer science keep fun in computing. When it started out, it was an awful lot of fun. Of course, the paying customers got shafted every now and then, and after a while we began to take their complaints seriously. We began to feel as if we really were responsible for the successful, error-free perfect use of these machines. I don’t think we are. I think we’re responsible for stretching them, setting them off in new directions, and keeping fun in the house. I hope the field of computer science never loses its sense of fun. Above all, I hope we don’t become missionaries. Don’t feel as if you’re Bible salesmen. The world has too many of those already. What you know about computing other people will learn. Don’t feel as if the key to successful computing is only in your hands. What’s in your hands, I think and hope, is intelligence: the ability to see the machine as more than when you were first led up to it, that you can make it more.’’

– Alan J. Perlis

• #1595340558 2020 Jul 21, 16:09 – Some resources for learning CS online

Check out Harvard’s CS50, teachyourselfcs.com and MIT’s Missing semester for some practical stuff as well.

• #1595283700 2020 Jul 21, 00:21 – Wealth shown to scale

• #1594743385 2020 Jul 14, 18:16 – IndieWeb

A fantastic wiki with a lot more pages than I initially expected! Click.

• #1594658110 2020 Jul 13, 18:35 – Startup Playbook

A wonderful resource I’ve found online: The Startup Playbook by Sam Altman.

• #1594631089 2020 Jul 13, 11:04 – Encrypted remote backups

Up until now I was just using rsync for backups, but today I wanted to start encrypting my backups. Initially I took a look into gpgtar etc., but I’ve just stumbled across restic, which looks like a promising solution.

• #1594561097 2020 Jul 12, 15:38 – Reimagining Vivaldi

I don’t listen to classical music often, but I do enjoy Vivaldi a lot. Here’s a pretty good recording on YouTube.

• #1594461065 2020 Jul 11, 11:51 – The \$25,000,000,000 Eigenvector

Some days ago I had to do some work regarding the PageRank algorithm for my numerical mathematics class and I was really surprised about how interesting it was. Take a look at this paper here. The math is actually not that hard; I’d say the ratio of complexity to effectivity is damn good.

• #1594051232 2020 Jul 06, 18:00 – I feel exposed by Paul Graham

• #1593775896 2020 Jul 03, 13:31 – Leveraging psychology when designing (digital) products

Here’s a website presenting several interesting psychological principles.

• #1593244110 2020 Jun 27, 09:48 – translate-shell cli translator

Works instantly without requiring an API key or any other setup. You can translate single words or whole sentences from/to different languages. link.

• #1593184591 2020 Jun 26, 17:16 – Understanding mathematics

A math professor explains how to approach math on http://www.math.utah.edu/~pa/math.html. Also, if this isn’t your typical math professors homepage, then I don’t know what is.

• #1593184545 2020 Jun 26, 17:15 – Some phlogs that could be interesting

Link: gopher://gopher.club:70/1/phlogs/. I’ve used lynx to briefly look at this list today and I haven’t really read anything off of it yet, but it looks promising.

• #1593184486 2020 Jun 26, 17:14 – Hello World!

Today I figured I’d like some sort of platform to post smaller snippets of information or links that could be interesting to others as well to, so I’ve quickly created this page. We’ll see if I’ll put it to use over a longer period of time…