Friday, March 27, 2015

[cxprlviq] EgyptAir 990

The NTSB investigation concluded that the crash was caused by the pilot steering the plane down.  However, the FBI criminal investigation found that that pilot didn't seem to fit the profile of someone suicidal.

Therein lies the hugely important puzzle which I don't think has received enough attention: it seems there are some extremely major holes in our understanding of psychology regarding suicide, things that, if understood, among other things (many other things), would be helpful in increasing air passenger safety.

We wildly speculate a mechanism:

Suicide occasionally pops into certain people's minds like a compelling itch to scratch, not consciously controllable and not obeying rational thoughts.  (Pretty frightening: anyone, including you, might suddenly off themselves one day and have no control over it.)

Certain risk factors may magnify whether thoughts of suicide turn into suicide carried out.

The speculated mechanism is infection by a microorganism, inspired by suicidal behavior in mice caused by Toxoplasma.  Our speculated microorganism also somehow benefits from its host dying, or at least did over the course of its parasitic relationship with the human species.  For example, it must spend part of its life cycle in soil, assuming burial of the dead has been common.

Any biological explanation of suicide must explain how it has persisted through evolution.  The reason humans haven't evolved resistance against this hypothetical pathogen is because the it taps into the natural ability of the brain to be intensely curious, to obsess, over something.  This capability as a component of intelligence has been generally evolutionarily beneficial.  The microorganism just slightly modifies that neurological pathway to make the brain obsess about death.

The effects of the infection (suicidal behavior) might also only occur later in life when the immune system has weakened, after children have been conceived, thereby avoiding evolution.

(Previously about parasites and evolution.)

Inspired by Germanwings 9525.

Thursday, March 26, 2015

[xuvlagam] Testing the false conviction rate

The justice system should regularly have tested its safeguards against false convictions. I suspect there are systematic failures.

Seek out tips and whistleblowers who know where the system might be broken.

Possibly use surveillance.

Wednesday, March 25, 2015

[ffotddzj] Notifying another search engine

Suppose you believe Google is currently objectively the best search engine but would like to voluntarily help another search engine, perhaps one making a better and more credible commitment to privacy, improve its results.  Create a browser plugin that records your searches with Google, the Google results, and the link you ended up clicking on, and sends all the data to the second engine.

Malware of course could do this, leading to the ethical dilemma: is malware to bootstrap an engine to break up the Google monopoly justified?

Could Google reverse engineer the plugin and poison the data? Probably need a web of trust.

[zreedqai] Front facing chess

The two players sit facing the audience, so they can see their faces.  They input their moves into laptop computers in front of them.  On a large screen above their heads is projected the board for the audience to watch.

[spoiajyo] Roko's basilisk

Censorship of Roko's Basilisk was an absolutely brilliant publicity stunt, absolutely consistent with their mythology.  I would not have learned of the Rationalists otherwise.

Does it represent a fundamentally new argument for censorship, or is it a mere application of the general principle of a benevolent entity believing some speech will cause you harm, so censors it from you as an agent acting in your best interest?

[zaarjrth] Streethenge

Create a program: Draw any line on a map (point and bearing) and the program tells you when (if ever) the sun will rise or set along that line.

Any astronomical object, any (or a given) angle above the horizon.

[sscborfc] Avoiding JPEG artifacts in PDF

ps2pdf -dGrayImageFilter=/FlateEncode -dAutoFilterGrayImages=false foo.eps

[xeplrmle] Tables

Hypothesis: a table is only useful as a format to depict information if it is small enough to fit entirely onto the display medium, e.g., one page or the screen or window.  This is in contrast to, say, lists, hierarchical lists, and even multiple paragraphs of text, which remain useful even if they span more than one page.

The profound corollary is that tables are only of limited usefulness: perhaps for information that is inherently small, or for pedagogy (an intentionally small example).  This provides a prescription of when to use tables.  And information that is inherently small can often be depicted in other ways equally well.

Tables do not scale.

Tables are the one HTML feature that Markdown supports poorly.  Perhaps this is fine.

A mathematical matrix is often depicted as a table.  However, if it is too big, perhaps it contains large algebraic expressions, or is a generic NxM matrix, it might be better to express the matrix elements as a list: a[i][j]=...

A counterargument might be that a large table is useful if interactive features are provided: row headers and column headers which stay on the screen while scrolling; ability to sort, select and hide columns.  However, this becomes more like the front end to a database than a table, and then we might want even more powerful ways, e.g., scripting, to interact with and  explore such large data.

Wikipedia struggles with large tables.

The inspiration was computer programming, in which there are often two-dimensional relations.  However, computer programs are rarely inherently small, so a language or IDE providing tabular features for programming is probably not useful.

[pvzhzttu] Two Schroedinger's cats

A single photon has its wave function split, and the two waves travel to opposite sides of the universe, eventually reaching two photon detectors each connected to a box stocked with a cat and poison.  The waves hit the detectors and cause both cats to enter and remain in a state of quantum superposition between dead and alive, the standard mind-boggling state of Schroedinger's thought experiment explored elsewhere.

But quantum physics is not yet done boggling your mind.  As soon as one of the boxes is opened, and the cat inside it is observed, say, to be alive, then the other cat on the other side of the universe instantaneously dies.  Perhaps because of the cosmological expansion of the universe, it is no longer even possible for a signal to travel between the boxes.  Nevertheless it seems the quantum entangled boxes have communicated faster than the speed of light across the gulf to ensure that only one cat lives: spooky action at a distance.

Sunday, March 22, 2015

[rsgtkngz] Bluetooth trackball

Operating a tablet with a Bluetooth mouse is going well, especially for clicking on the tiny Xs to dismiss obscuring advertisements on webpages.

Given that tablets are often used in "comfortable" places without a hard flat surface on which to place and operate the mouse, a Bluetooth trackball would be even better.  But no such device exists.

This would be a use case for a hypothetical USB to Bluetooth converter.

[rcfbyrhl] Velcro

Velcro is a nice technology, but deployed less than I would expect.  For example, the ancient technology of laces still dominates shoes.

Is it purely because velcro is not considered classy?  But the prohibition seems to universal to fashion across all classes.

Velcro is a more accessible technology, usable with one hand, requiring less fine motor control than laces.

[nabpwujy] Parents not abandoning their children

In a completely anarchic society, what incentives are there for parents not to abandon their children?  In particular, both parents abandoning a child before he or she can fend for himself or herself.  The inspiration is that there do seem to be anarchic societies, yet they don't seem to be suffering drastic population decline due to child abandonment.

Hypothesis 1: The biologically hardcoded behavior of parent-child bonding, especially mother-child bonding, is enough.

Hypothesis 2: The biologically hardcoded behavior of parent-child bonding is not enough.  However, this combined with the observation that there doesn't seem to be vast amounts of child abandonment anywhere leads to a profound corollary: there are no true anarchies.  Every society, even if it does nothing else, finds ways of making parents behave properly, perhaps through informal regulations enforced by the extended family.

If hypothesis 1 is true, then laws forbidding abandoning children are redundant and unnecessary.

If hypothesis 2 is true, then the situation is more complicated.  We predict that if such laws were repealed, then society will naturally develop informal mechanisms to accomplish the same task.  However, those natural mechanisms might be to reenact the laws.

[eiqyedzx] Tamora responds

After the revelation of cannibalism in Titus Andronicus, some possible surprising alternative responses:

In my culture, the bodies of the dead have no special value; kudos to whoever thought of this and to the chef for making them tasty.  Is there any more meat left?  There was a lot more of them than what I just ate.

Never liked those kids anyway, and they could have caused succession problems.  Good riddance.

I always knew I was eating long pig; I have had it often before.  It has been the taste of victory and makes me happy.

Saturday, March 21, 2015

[qgvqsehi] Should have known

An employee does something bad and the employer is held legally responsible for the damages caused.  So far, nothing new.  The employer also suffers damages in reputation in the court of public opinion.  Also nothing new.

If the employer knew (or could have known) via a background check that the employee was prone to causing that kind of harm, is the employer liable for more than just the actual damages?  Perhaps "gross negligence" in the employment selection process?

If so, this creates a problem: a risk averse potential employer has even more incentive to avoid hiring someone with a bad background, creating a de facto blacklist that no one is willing to employ.  By statutorially setting the amount of damages for such "gross negligence", society can enforce the blacklist as strong as it wants.

The most prominent such blacklist is the sex offender registry: is the employer liable for gross negligence for hiring someone on it? There is almost certainly political support for such a regulation.

[ttzjifqs] Wrongful deaths by the military

Consider making it possible to invoke, and win, criminal and civil legal proceedings against harms caused by military actions in the absence of a legislative declaration of war.  It stresses the importance that the people control the military.

Surprise military actions are possible by retroactive declarations of war.

Who is liable, the soldiers following the orders or the commanders giving the orders?

[vhryptvn] See all around you

Create a first person perspective game in which you can always see 360 degrees around your head, yet projected on a standard rectangular monitor.  No enemy can sneak up from behind.

Which is less disorienting?  A 180 degree front view and a separate 180 degree "rear view mirror", or a single 360 degree view wrapping all the way around?

An even greater challenge is to display an entire sphere around you while minimizing distortion (strictly an impossible task).

[pywhlgds] Quarter area mover

Some fairy chess piece possibilities involving being able to move to a relatively large area.  The goal is to lessen the relative weakness of pieces that move in a line as the board area increases quadratically.  (Though a piece that moves in a line can reach quadratic area in two moves.)

The piece jumps to any other square on the quarter of the board it currently occupies.

The piece jumps to any square in the rectangle bounded by its current square and the nearest corner. The closer to the center of the board the piece sits, the more powerful it is.

Both of these pieces will need some additional kind of move to escape the quarter or corner of the board they are currently on. 

Other possible variations such as color bound, defining the target area by an edge instead of a corner.  Can only move to a subset of the target region reachable by some (perhaps unlimited) number of micro steps of certain types.  Limiting to two micro steps is a hook move.

[gfkpocuq] Epilogue: Beyond Good and Evil

Good has finally defeated evil.  We pick up where the original story ended: yin and yang have thus become profoundly unbalanced, causing unexpected destructive consequences.  Perhaps a reevaluation: that which had been labeled Good or Evil is not so simple.

Harry Potter
Star Wars

[clqucfzx] Rational jumper to edge

An "edge-rational" nightrider fairy chess piece can repeatedly jump at any rational slope subject to the constraint that among the squares on its path extended must be one of the squares on the edge of the board.  Equivalently, it can jump to any edge square, but also stopping at squares on the way to the edge which are at the same slope.  It can capture or be blocked by pieces on those intervening squares.

The edge-rational knight is like the nightrider but cannot repeatedly jump; it stops at the first square of a permitted slope.

On average, how many squares can an edge-rational nightrider reach in one move on an empty board?  Knight?  What is the asymptotic behavior as the board size increases?

Where can it go in two moves?

Its behavior is number-theoretically interesting, especially if it is a prime number of squares away from an edge. It can always move like a queen because 1 divides everything. Perhaps disallow those moves.

Could allow or disallow moving backwards along a path to the edge square.

Rather than defining the edge squares the region its path must "hit", we could instead define the target as some other subset of squares, e.g., just the 4 corner squares.

Could make the target the squares one square beyond the edge. This eliminates always being able to jump to any square on the edge.

Thursday, March 19, 2015

[pqvfpeme] Interleaving error correction

An error correcting code expands data blocks of size P to larger blocks of size Q.  The sequence of N such blocks stacked on top of each other can be visualized as an N by Q matrix.

Transpose the matrix then serialize the data by rows.  (Equivalently, read the original matrix by columns.)  This is known as interleaving.  Interleaving causes bursty errors to be spread out among many blocks (which are then likely recoverable), instead of unrecoverably affecting a single block.

Transposition of a very large message can be done using a cache-oblivious algorithm.