Tuesday, November 25, 2014

[ntbmlntu] Recursive image decomposition

We can divide a rectangle orthogonally into two rectangles by specifying the position and axis of the cut.  This can be repeated recursively.

Given a rectangular image, is it ever beneficial for image compression to cut it into pieces and compress each piece separately?  Assuming so, given an image, determine the cuts which maximize compression.

There is an additional subtlety of the order the pieces get compressed.  Older pieces may affect the context of neighboring newer pieces across an edge.

There seems a similarity between this decomposition of space and decomposition of layers.

[esksbznu] We require more hydrogen gas

Construct a star out of hydrogen gas.  Keep adding more hydrogen.  Does the radiation pressure from fusion always balance gravity (while hydrogen there remains hydrogen)?  Or, with sufficient mass, can a ball of only hydrogen directly undergo core collapse?  Maybe this was possible in the early universe.

What if you start with some other element?  Do all elements lighter than iron essentially act the same as hydrogen?

[iwhehdmx] Alpha decomposition of an image

Suppose an image is constructed of one image layered over another with alpha transparency.  Is it ever beneficial from an image compression standpoint to compress the layers separately?

Assuming it is, consider the much harder reverse problem: given an image, decompose it into layers to maximize image compression.

Monday, November 24, 2014

[zhnyxvye] Lzma contest

There seems to be many different compressed inputs that an LZMA decoder will uncompress yielding the same file.  This suggests a contest: find the smallest file that will uncompress to a given target using a standard LZMA decompression program.

[zhnyxvye] 3 person chess match

A three person chess match has the feature that the audience will get a game everyday, but the players will enjoy a rest day every three days.

[uvmjqhcm] Automatically deriving the rules of chess

Given a collection of chess games, how difficult would it be to automatically derive the rules of chess?

Consider a relatively difficult input format: a game is presented as a sequence of [0..7]^4 tuples (i.e., 12 bits per half-move), representing start and end squares in the standard way, including castling represented by extended king moves.  Not even the result of the game, e.g., 1-0, is provided, but we assume the games have been played out until the result can be determined mechanistically from the state of the board (no resignations, forfeits, losses on time, draws by agreement).

The computer must derive not only the rules but the objective of the game.  The example games must have been played by agents seeking to win, perhaps loosely defined as never missing mate in N for some small N.

Perhaps this is not enough, and the computer is also provided afterwards with an interactive phase in which it can ask questions about positions not seen in games: "is the following move sequence legal?"

It is hard to imagine that a computer will ever be able to figure out on its own the prohibition against castling through check solely by looking at games, but it could be possible: permitting castling through check would have provided a mate in N or a way to escape mate.

One motivation was laziness: specifying the legal moves and the objective of the game are probably the most error prone and tedious parts of any game programming. Doing it automatically, even semi-automatically, would be great.

The other motivation was considering playing correspondence chess with distant aliens. How can one communicate the rules of a game to the alien culture with which one shares no language?

[bxjumjdl] Sex never exceeds expectations

While it is easily possible for an act of sexual intercourse to be worse than expected, can it ever be better than expected?  If your gut feeling is, sex with this person is going to be unpleasant, is that gut feeling ever wrong?

The hypothesized answer is "no", by the following mechanism: the pleasantness of sex derives from the one's mental model of the other person. Great sex must engage the mind. If one's mental model of the other person is not appealing (which is what your gut feeling reports), then sex is doomed not to be very pleasant.

The only way this could be wrong is if one's mental model of the other person changes, but we disregard this possibility as unimportant because we assume there has already been a "getting to know you" phase to acquire enough information to form a mental model accurate enough to evaluate having sex.

[ksgayynq] 15 or 16 go

Elegant is a go 囲碁 board of size 2^n or 2^n-1, which can be recursively divided into quarters.  There is a center cross for 2^n-1. Coordinates may also be written in binary.

[jbyztngl] Olbers paradox brightness

If we lived in a universe in which Olbers's paradox were true, i.e., infinite universe, no redshifting, no big bang, then how bright would the average night sky be?

There would likely be deviations from uniformity of nearby bright stars.

Sunday, November 23, 2014

[uasebrpb] Parrises Squares

Invent a future sport which causes massive and traumatic injuries by today's medical standards but assumes (perhaps humorously) that medicine of the future will be able to treat such injuries easily.

Dodgeball with lightsabers: "It's just a flesh wound."

Particularly creepy by today's standards might be a player suffering traumatic brain injury which is "healed" by removing the brain and inserting an artificial one containing, or running like a computer program, a copy of one's consciousness from before the injury.  The original consciousness experiences dying a horrible, painful death.

Saturday, November 22, 2014

[qocvmftl] Quasars versus intelligent life

Assume all civilizations hunger for more and more energy, so they will all naturally gravitate toward the most powerful compact power source available: a rotating supermassive black hole, extracting rotational energy from it perhaps by the Penrose process.  In this way, they will all become Kardashev Type 3 civilizations.

If this is true, then we would expect that most galaxies to already have their supermassive black holes tapped out. If it is possible to tap such black holes, then we would expect the technique to have already been independently discovered and deployed countless times all across the universe.  We should not expect to be so lucky as to be seeing a quasar or active galactic nucleus before some civilization taps it.  Both quasars and AGN are thought to be powered by the rotational energy of supermassive black holes.

Then, what could the fact that we do see quasars mean? Here are some possibilities:

  • There is no intelligent life in the universe, where "intelligent life" means type 3.
  • The universe is still too young too see significant galactic and intergalactic conquests.  This is especially true for distant quasars which we observe in from the early universe.  Perhaps type 3 civilizations occur extremely rarely, far less than 1 per galaxy, and intergalactic travel, needed for the sum of such civilizations to tap most of the supermassive black holes in the universe, is so slow that that it hasn't happened yet.
  • Quasars and AGNs are flukes that are perhaps weird or temporary that most type 3 civilizations don't find it worth the effort to tap when their are plenty of other "normal" supermassive black holes out there.  Perhaps they are galaxy mergers.
  • Extracting energy from supermassive black holes is always so much more difficult than, say, capturing the energy of all the starlight in a galaxy that it is never worth it for a civilization to do.  I suppose this is a dark matter theory: there exist dark galaxies, a tremendous number of them, whose light is being mostly harnessed by a civilization, but they still do (of course) show up when accounting for mass.  But I think we would still expect to see their heat output.
  • There exists some other power source that all civilizations discover that makes tapping supermassive black holes not worth the effort.

[ywofedam] Supermassive black holes shouldn't rotate

Supermassive black holes grow by swallowing other matter.  However, due to frame-dragging, a rotating black hole will preferentially swallow matter approaching it at the wrong angle, against its rotation, so with the "wrong" angular momentum, so swallowing it would cause it to rotate slower.  Matter approaching it with the "right" angular momentum will be more likely to receive a kick in velocity due to frame dragging, and that kick may allow it to escape being swallowed.  Also, that kick will consume some of the black hole's rotational momentum also slowing the black hole's rotation. (Similar to the idea of dynamical relaxation causing evaporation.)

This ought to result in supermassive black holes with almost no rotation.  But quasars are thought to be powered by a supermassive black hole's rotational energy (Blandford-Znajek process).

[uydztnvm] Artificial Berlin

Consider a simple chess variation in which the initial position is altered by removing both queens.  This could be easily incorporated into orthodox chess by some some initial knight maneuvers.

Inspired by the Ruy Lopez Berlin endgame.

In popularizing the Berlin Defense, showing it can withstand Kasparov, probably the greatest of all time (in opening theory, at least) under championship match conditions, Vladimir Kramnik may go down in history as one of the greatest influences (for better or worse) that shaped how the chess opening is played.

[bqrkysqw] Manipulating the press

A female game developer allegedly tries to manipulate the press and all hell breaks loose.  Uber allegedly tries to manipulate the press and all hell breaks loose.

The common thread is the parties feeling it worth it to attempt to manipulate the press.  This suggests a failure within the journalism institution itself.  In order to maintain credibility, one would expect journalistic institutions to have developed powerful mechanisms to avoid manipulation.  And the market rewards discovering and publicizing if a journalistic competitor is being manipulated by whom and why and how.  These forces should be making it so difficult to manipulate the press that no one should even be trying.

The one exception is government: a government has so much power that no private journalistic institution would be able to resist government manipulation, which is why we have explicit freedom of the press.

Counterpoint: maybe readers don't care about the credibility of their journalism much, regarding journalism not as a source of truth but entertainment.  In which case we would not expect journalistic institutions to have deployed powerful mechanism to avoid manipulation.

[zcyhopvf] Transcription to text figures

The existence of text figures (lowercase numbers) suggests a lossless encoding of text to numbers, preserving punctuation and capitalization. Use the straddling checkerboard.

It is not intended to be a strong code, but rather artistic.  Perhaps to make text easier to OCR.

We need some escape characters to encode numbers themselves.  The straddling checkerboard provides 2 (8+10+10-26) which is just enough for "The next character is a numeral" and "The remainder of this word is a number".

Friday, November 21, 2014

[xneezhkh] List or cons cell

In Lisp, despite its name, the fundamental data structure is not actually the list, but actually the two-element cons cell.

Should this be regarded as a fundamental feature or an artifact of implementation?  In praising the simplicity of Lisp syntax, I usually ignore this artifact, as it introduces one more lexical atom: the period.

It seems anything that can be done with a cons cell can be done with a two element list.  I suppose a cons cell automatically enforces that the length of the list is exactly 2.

[oiqhiqrl] gpg passphrase hint

GnuPG needs an additional field in private keys in which to store a passphrase hint.

While usable in the standard way (possibly weakening encryption), the way I would like to use it is to store the parameters to an external PBKDF algorithm (e.g., scrypt) which can exceed the 65 million iteration limit that standard GnuPG imposes in its S2K key stretching algorithm.

Wednesday, November 19, 2014

[ltykhuqq] Volcanic mountain ranges

What mountain ranges are composed entirely of material from volcanic eruption, as opposed to uplift? Inspired by wondering if every hill in the Cascades is an eroded volcano formerly on par with the current tall ones.

Obviously Hawaii.

Sunday, November 16, 2014

[fhhksoph] Star Wars Redemption

Darth Vader's conversion back to the Light side reveals a profound horror to the Jedi: they had always been too quick resort to violence -- murder -- to deal with the Sith.

Wednesday, November 12, 2014

[rjbnzsca] Suffrage 2nd order cynicism

The standard cynical explanation of why women were granted the right to vote is it was purely means to power by political forces which inherently care nothing about human rights but only political power.  They invoked the standard "you are oppressed; we will make things better" message (or lie) to mobilize political action.

However, if this explanation is true, why wasn't it successful earlier?