Friday, January 31, 2014

Friday: Did everyone forget how to draw a stupid free-body diagram?

Part 9044 in the never ending list of "things I find on the internet and then get grumpy about."  This thing:
I found it here.

It's not magic.  Let's consult an expert:

Thanks, PB.
So I chopped the problem into the identical quarter problem, assumed we only need to deal with the vertical motion, and pulled out my "you learn this in the second physics class you ever take" abilities:


How does it work? It's suspended from the hooks in the ceiling, duh.  How much weight can you add to the bucket before it falls down?  However much you want, as long as you don't increase the tension in the string so much that it snaps.


I have found a nemesis

Martin Short, assistant professor of mathematics at the Georgia Institute of Technology in Atlanta.  You are now my nemesis.

Why?  Oh, just because of this.

I totally had a quick mock up simulation with agents and cops to simulate crime rates, and you come along and do it all "better" and "in two dimensions" and "with a youtube video".

I was going to put that in a blog post in the future.  Now what will I do?

Yes, fine.  I'll probably post a picture of a panda and talk about sushi.  Still.  It's the principle of the thing.

Or maybe chirashi? I haven't had that in years.  

Thursday, January 30, 2014

Thursday: Draw a Dinosaur Day!

I went with a stegosaurus this year.  I made it a wizard by putting a hat on it.  The sun is a wizard too, because I didn't want it to get jealous.

Wednesday, January 29, 2014

Wednesday: I need a tag for images that I'll likely want to use again in the future.

That would save me lots of time trying to find those images over again.

First though: I forgot to post this Pikachu from Saturday:
There's a dancing banana too, but that wasn't the point.  The shrimp udon is kind of cute, too.
Anyway, I saw this link on fark about how mayo is "king of condiments," not ketchup.
If you look at their data, it looks like mayo has lots more than ketchup.  However, if you look at the axis, they're plotting "billions of dollars."  I don't know about you, but I rarely buy my condiments in increments of dollars.  I buy fucking bottles.  Except mayo.  I don't buy that.

So how much does a bottle cost?  Let's have google look that up for us.  I used real brands that have comparable name recognition.  To me.  Plus, these all nicely have various size options.


I don't have their actual data (apparently from this source, which I'm not about to spend $900 to actually read), so I pulled up the plot in the digitizer, and it tells me that in the last year of study, mayo sold about $2e9, and ketchup sold about $0.8e9.  Now pull out the stats, and assume you're buying the cheapest option per ounce.  This gives ketchup at $0.079375/oz and mayo at $0.18136/oz.  This translates to ketchup and mayo selling $1.0079e+10 and $1.1028e+10 ounces each.

"See, mayo still sold more, therefore is king."  Fine, that math didn't totally work out the way I expected.  However, think about how long these things last.  Consumer Reports says you can keep ketchup for six months, but mayo is only good for two.  I can't imagine a consumption model based on those facts that doesn't suggest more mayo is wasted than ketchup.  It doesn't need to be that much more to make up that difference in sales (10% more wasted mayo than ketchup will satisfy that).

Next, McDonalds.  They're not paying retail price for their ketchup, so big consumers like them are going to make the actual sales of ketchup skew higher than this estimate.

Finally, mayo is disgusting.  QED.


Tuesday, January 28, 2014

Tuesday: Hey, I know where that is!

I saw this news story before I left this morning.
This is from the hall outside my apartment.  That school's about a third of the way from the top, just left of center.  Also: helicopter in the upper left corner.


Monday, January 27, 2014

Monday: Three meetings today

I kind of want tomorrow to be the weekend again because of this.

Was nothing in the 50's not terrifying?

  • What have you done, Japan?
  • I'm not a fan of Hal Jordan.  He's basically the absence of personality.  At least I'm assuming this is Hal.  In any case, I don't really like Hal.
  • This is why gun control is necessary.  Because some people are too old and terrified of the world to be trusted with the ability to hurt people.
  • I like burgers.  I like monte cristos.  This is an abomination.
  • "1000 years! REFRIGERATOR!"

Sunday, January 26, 2014

Sunday: Burger day

I spent a large part of the afternoon trying to decide what I wanted for late lunch, eventually coming to the conclusion that I wanted to try to make this burger again.  I didn't do the bison meat for this one, as the flavor was lost last time.  I also made the mistake of adding sprouts.  That wasn't a good idea.
It's free "double sauce" month at the Counter, so I tried the peanut sauce.  It wasn't very good.

And they kind of burned my cheese fries.

Saturday, January 25, 2014

Saturday: Not really an adventure.

Today's plan was to go get sushi, then see a movie.  How hard can that be?

Turns out it's also "sidewalk sale" day at the mall.  Sidewalk sale day is the day that all stores put things they would be more ok with being stolen outside the store, on sale, and then put a person to watch all those things, because although they may be "more ok" with it, they'd really prefer nothing get stolen.  

It also makes going to the mall an exercise in not killing everyone, because the fourth time you nearly run into someone who has decided to come to a sudden stop to look at some shit on a rack that's turned the ready cramped corridor into a fire hazard and now I'm so far into this sentence that I've totally forgotten how I started it, but I don't care, and as a side point, I'm not sorry I called you "in the fucking way," because you were.  In the way.  The fucking way.

I should have known there was going to be trouble when the exit ramp was backed up.  Onto the highway.  Or, to be technical, onto the shoulder of the highway, because some moron decided that was the way to fuck up traffic the most.

In any case, I went to put in my name for sushi, only to discover that with the current wait (30 minutes) and the current time (1:15), I'd be unable to make the movie (1:50).  That meant I had to switch to the backup plan, which was to wander the mall to kill thirty minutes, and then get some theater food.

Remember when I talked about Thinkertoys, and the people getting arrested there?  These are the "goth jigsaw puzzles" I was referencing.  I am completely baffled by who would buy these.  I keep coming up with "time travelling kids from 1992".
 Then I looked at the cooking ware store, which really has oddly specific things.
This seems unnecessary.  You'd have to underfill them to make sure you didn't get big puffy muffin tops, so you're left with tiny kitty face muffin butts.  Plus, you'd have to make like 99 batches with most recipes.

Now this I would have bought if I didn't have a day of not wanting to carry a package around already planned.  This is obviously something that has a thousand uses.
Then, due to this link that I mentioned on Tuesday, I went to go see Frozen.  The short review is that this is the best Disney movie I think I've seen ever.

Looking at this list (and excluding the Ghibli movies, as that's just cheating), I don't see anything else that really even comes close.  A lot of the reviews I've seen talk about "oh this is as good as the Little Mermaid, and Beauty and the Beast."  No.  Those aren't bad movies, but not really this level.  The characters in Frozen actually seem like real people, not some sort of caricature of how people act.

Of the points in that link, I'm on board and happy with all but the first one.  Yes, they panic and isolate Elsa, which isn't really a great idea, but then they die (spoiler, not spoiler.  This is a Disney movie.  How many of those don't have dead parents?  Mulan and B&B.  I almost added Lilo and Stitch, but she's Lilo's sister.).  In the absence of any new orders, everything just kept going the same way.

I continue to not unerstand why people leave before the end of the credits.  Pretty much every movie has secret scenes now.

Anyway, everyone should see this, and now I see why the internets have been talking about it for three months.

Then sushi.
Plus ticket stub to make this not just the same boring sushi photo.


Friday, January 24, 2014

Friday: Links first, because I have like 6MB of Oshawott gifs

  • Pokemon.
  • Cartoons.
    • I think it's reasonably clear that these are grown up PPG, but I'm still surprised someone didn't realize that this was a bad idea before it hit the public.  Since, you know.  Kid's show.
    • I need to catch up on my Adventure Time comics.
    • I wonder how much shit this Ghost in the Shell movie will be.  I'm guessing painful CGI Tachikoma, and a complete misunderstanding of the themes, leading to a generic action movie.
  • Other stuff.
Now a cut, in case you don't want to look at Oshawott or download a bunch of gifs.

Also, this nice picture of Pikachu.

Thursday, January 23, 2014

Thursday: Isn't this an obvious thing?

Why don't the rich consider themselves rich.  That link has a whole bunch of bullshit associated with it.  Now, look at this link, which has a simple income percentile graph based on census data.

There's a curve there, and I'm going to be lazy and assume this curve has the form of a logistic function:
p = 10**(1 / (1 + exp(- 3.5 * (log10(x) - 4.2)))) - 1) / 9
See, I didn't just make that all up.
It's not perfect, but it's close enough for the next step.  Assume people hang out with people who are roughly randomly sampled around their own wealth percentile.  Basically, everyone hangs out with people that are "in the same income group" in some fashion.  However big you assume that box is, your income relative to that of the wealthiest person you know is basically the inverse derivative of this curve:
I switched to log10(income) because that simplifies the math.
See?  If you make $1e5 a year, you probably know someone who makes 10**(0.75) = 5.62x more than you.  You make a million?  Oh, I'm sorry, someone you know makes a hundred million, you poor asshole.  

Yes, my initial assumption that people hangout in percentile cohorts is poorly supported.  However, this is probably easiest to support, especially in the highest ranges (5.5 and higher), as we're starting to talk about high paid doctors/lawyers/CEOs.

Also, there's a feature where poor people realize that they're poorer than everyone else they know.  Unfortunately for them, this isn't really incorrect.  They really are really poor.

Wednesday, January 22, 2014

Wednesday: Better than garbage with one algorithm

It's clearly not perfect, but my implementation of the first pass (and obviously simplest) algorithm tells me that the value that should be ranked first comes up with a rank of second.  I didn't do a full study of the errors, but a few spot checks indicate that I'm not completely crazy.  Good to know.


I decided today to convert some of my leftover tikka masala into pizza form, using a pita and some fontina cheese.  This was a good idea, and I believed in myself, and so made it a personal pizza.  I'm also glad I got the $6 pitas and not the $3 pitas.  These had a nice flavor, and weren't bland styrofoam like a lot of the other ones I've had in the past.


  • I saw this article earlier today, and a quick read through brought up the following points:
    • K-modes is basically just K-means, which is kind of the brain-dead-iest way of saying "put things into groups, automatically."  Nice shout out for algorithms, but it's not really a breakthrough.
    • The clustering should really be a product of the webpage.  They have this damn data, why aren't they doing this analysis?  Similarly, ranking questions by applicability should be the default.
    • "so he wrote a new program to visit the pages of his top-rated matches, cycling by age" This is where things go off the rails from "should have been already implemented" to "kind of creepy."  This is kind of similar to ringing a bunch of doorbells, and then seeing who answers the door.
    • "It was first date number 88." Ok, so, great that the method told him how to get sufficient attention for a first day, but this still kind of boils down to Monte Carlo sampling.  That's marginally better than 1%.  If algorithmic improvements really only boost the effectiveness to 1%, that's a really shitty system.
  • Next up: the new Powerpuff Girls special.
    • I didn't like the new art.  I especially felt that the first five-ten minutes were over-exposed, washing out a lot of the color and detail.  It did improve towards the end of the episode.
    • The writing was good.  This wasn't "last season of Dexter's lab", where everything is different because the new team doesn't understand the characters.  It looks like part of the reason is that the director also wrote some of the old episodes.
    • It's good to see Ringo still getting work.
  • Wonder Woman.
  • Finn and Jake in the best Jake transformation scene ever.
  • Grover.

Tuesday, January 21, 2014

Tuesday: Mathmotrons

Because of this news story, I'm now suddenly interesting in solving a math problem something about sports.  Whatever, in any case, sports is all just sampling from a set of possibly unknown distribution functions.  So as long as you come up with a plausible simulation of those distribution functions, you can "do the sports" without having to bother with all the time and effort.  Sure, rooting for the New York RANLUX against the Denver Tausworthe's would be silly, but really, it's basically the same thing.

Anyway, I'm going to do math at some historical data to determine how well my math can math, and then see if my math can math it better than other people who actually care about the "teams" and "real human beings".

I don't have any squirrel pictures for today.

Monday, January 20, 2014

Monday: MLK day

Which means I get the day off because my state isn't horrible, and that means I went out to get sushi today.
You might not be able to tell, but this was a beautiful piece of salmon.
Oddly enough, for absolutely no good reason, I realized that going for sushi involves two Poisson process problems that should be solvable, but it's 9:30, and I want to get this post up and done by 10, so it might have to wait until tomorrow.

The first problem (which I think is probably easier) is the Counter Problem: Given a 20-seat counter (because that's how long it is), is it always best to seat in party order, or can you minimize the amount of waiting using some other method.  Assume all parties have between 1 and 4 members, and that each party takes k minutes to finish.  Minimize the person-minutes of waiting customers, who arrive in parties every m minutes (where 20 / (avg(party size)) = m / k for a steady state solution).  This should be easier because it's basically a generative process:  make the distributions, let it run, and see what you get.  Blammo.

The second, more interesting, problem is what I'm calling the Karaage Problem:  Given an observed set of d dishes of karaage on the belt, estimate how long they've been out.  The motivation here is that fresh karaage is better than old, so you don't want the one that's been sitting out for 45 minutes.  The assumptions here are pretty simple:  karaage always comes out in k dishes because they fry them all at once, and they're then removed from the belt by a Poissonian process.  There's a meta-solution as well, which tells you how old things should be if all karaage is drawn by people who know the solution to the original problem.  This is the one I really want to know, because I think k is equal to 5 or 6, and there were d = 3 today.  I didn't take one, because my guess solution meant this was like 20-30 minutes, and that didn't sound great.


After sushi, I went to get groceries, luckily finding that after two weeks, pastrami is available again.  Why was there a run on pastrami?  Who knows.

In any case, I got the idea that I wanted to get stuff to make tikka masala, and have that this week.  I picked up a packet of this sauce stuff, and some naan.  I forgot to take a picture of the final product, which is why I completely forgot about this until I found the link in the linkpile.  My conclusions are:
  • This is not as good as actual restaurant TM, and if I actually wanted to spend effort, I could have cooked a better one from scratch.  Probably.
  • This really only takes like no effort to make, so the tradeoff is marginally worth it.
  • It calls for two pounds of meat (or veggies), and that's probably about right.  However, it looks like it doesn't make too much, probably because there isn't a bunch of extra sauce.
  • WF brand garlic naan has a really weird metallic taste that I don't really like.

It's like the entire internet opened up, made something, and then dared me not to share it.



The fact that the official CN Africa made this blows my mind.  It does help explain the fact that those are totally new poses that were animated for Marcy and Finn (at the very least).

Sunday, January 19, 2014

Sunday: Adventures?

So guess where I haven't been in like four years?

Yeah.  This isn't as cool as the Academy of Arts, as it's more about "Hawaiian history" than "all the arts. All of them."
They did have this delightful tiny turtle, sitting on a tiny hand mirror.  The phone camera kind of donked it up, but this thing was super crazy detailed.  "Turtle Toy."  Yeah, no, explanation card.  You don't give that to a kid to play with, even if they are royalty.  Probably especially if they're royalty.  That thing says "beautifully carved choking hazard" in like seven languages.
 Then, I discovered that you can't turn left onto the Likelike.  Oops.  Important other note for my future self: when you get to the Kaneohe end of the Likelike, you don't want the exit you think you do.  Due to reasons (I suspect massive head trauma), the exit on the right doesn't head east like you'd think, as it crosses (at grade. Again, head trauma) back over the Likelike and takes you to that one intersection you've memorized as it's the one to get from the wrong road to Windward mall, and therefore the right road.

Anyway, if you clicked through to the map, you can see what the final destination was.  Cafe Sistina, which I tried last year.  Part of the reason I didn't abort the excessively long trip was that it was a bit early for dinner, and I didn't want to get there before the kitchen was ready.  In any case, I still ended up being the first person there.

I think this visit clinched this place as "looks nice, but it just seems like the food isn't up to the right level."  By the end of my meal, I think I narrowed a significant issue to a lack of salt.  That's the only thing I can think of to explain why all of the dishes I've had here have a muted, drab flavor.  If there isn't enough salt, the flavor doesn't pop out as much, and you get this.  I understand "you can salt things yourself," but no, that's not the way to cook things.
 "Bruschetta Pomodoro 6.50 Grilled bread rubbed with garlic, topped with chopped tomato, basil, extra virgin olive oil"
I wanted something light to start with.  The olives were way too strong for the rest of the dish.  Not bad, so maybe I don't hate olives as much as I think I do.  The cheese was a very sticky crumbly one.  Almost like a feta, but with a taste like a goat-ier mozzarella?  Any guesses?

"Meat Lasagna 14.75 Sistina's special layers of pasta, roasted lean Meat, tomato sauce and cheese, ( no cream )"
This was the reason I wanted something light.  I had intended to do lasagna.  So.  Um.  Yeah.  Identify the problem here.  Did you say odd Parmesan distribution?  Sure, I could be picky about that, but no.  It's the burnt top pasta sheet on the right, and the dark-colored right side.  I know I don't run a restaurant, but isn't this the kind of thing that you cut off and not serve?

This makes me think that there isn't really anyone checking dishes before they leave the kitchen.  That would also help explain the lack of salt.  No one's checking sauces and things to be sure they're correct, and also not checking that things that really shouldn't be burnt aren't burnt.  The rest of the lasagna wasn't too bad, although again, a bit drab in flavor.  And this is my example of salt needing to be fixed in the kitchen.  How do you fix the salting on a dozen layer thing?

I was also down on the vegetable blob there.  Zucchini/onion/bell peppers stewed until mushy and served on the side.  I poked at it, but didn't really try it.  Miscellaneous mushed veggies isn't really my favorite, and I know I don't like zucchini.

They do have good art, even if I can't get it to be in focus.  I'm positive this looked better on my phone screen.

Dessert:
Panna cotta with strawberry/balsamic sauce.
I was hoping for chocolate cake, but they were out.  This was pretty good, not at all a "milky jello."  One slight issue was that the strawberry sauce was a bit too sweet, which pretty much cancelled out the vinegar bite.  This, less sweet, with a tiny dish of very coarsely ground black pepper?  That would have been even better.

So, I'm down an Italian place again.  Verbano, maybe?  The King street one closed, but there's one in Kaimuki (apparently right next to the Mexican place?) and another in Aiea (that was better than the King one anyway, if I remember correctly).


  •  After reading this interview, I want to beat Glenn Harlan Reynolds with books.  
    • Higher education prices are rising because government funding for those schools has been slashed to almost zero, forcing the costs onto students, who are then taking loans out from the government to pay for things.  Yes, that pushes the debt burden from "the country as a whole, with a perfect credit rating" to "fucking college kids, with shitty credit."  If you can't see why this is more expensive, you're a moron.  
    • Homeschooling and online courses.  Go fuck yourself, Jimmy McDoesn'tNeedTwoIncomesAndHasWonderfulResources.  Go get a shorter name, too.
    • School choice.  Because if a school is failing, we should just close it down and move everyone elsewhere?  When did everyplace magically have the resources to fund multiple redundant schools?  If you don't increase funding (which would probably solve the problem anyway), you just force more kids into an overloaded "good school," probably detrimentally.  Oh, what's that?  You'd make sure to keep the good school good by not overcrowding it?  So now you're just keeping kids in the bad school?  If you have an explanation for how this works that doesn't involve doubling funding or pulling shit from your ass, I'd like to hear it.
    • "You have the quasi-monopoly with mostly unionized teachers in K-12 and the tenured professoriate in higher ed." Then explain adjunct faculty, you absolute shitbag.  Go ahead.  I want to hear you fucking explain that.
  • I didn't see Lilo and Stitch until I moved to Hawaii.  I would totally watch this sequel.
  • Well, now I want nachos.
  • This was unexpectedly depressing once I realized that who Eustace was.


Saturday, January 18, 2014

Saturday: fishes

It's a long weekend, so I decided to break with tradition and get different lunch than Saturday Sushi.  I still felt like fish, so I decided Nico's would be a good choice.  I was a bit lazy, and nearly lazied myself past the end of lunch (they stop serving at 4).

It's a  reasonably good time, though, as there aren't many people there.

Standard option.  This was mostly what I was looking forward to.

I also got a scoop of ahi poke for dinner.  I had planned to have this on top of a bunch of arugula, but my arugula smelled off when I opened it, so I guess it's gone off.  

Friday, January 17, 2014

Friday: Dude, that's a bad idea

I mean, that's clearly a skeleton with a face strapped to its head with a skull mask over the face.  He's not a very sneaky skeleton.  You're some sort of priest or bishop or something.  You can dispell that guy.  It's like you haven't read the PHB at all.  You've got a damn polearm too, that has to do at least 2d4 damage.

  • Japan. Not giving any fucks since forever.
  • Fictional Japan. Not giving any fucks since forever.
  • Cats.  Not giving any fucks about property rights, because they're goddamned communists.
  • Pokemon.  Sometimes giving fucks, other times actively removing fucks from existance.
Serious links.
  • Damn it, animal planet.  I'd like to make a fantasy puppy bowl team, except I can't because it's run through facebook..
  • I saw the original video, and nearly made a blog post stating "this is bullshit, because I'm pretty sure infinite series don't work like that."  However, I sat down, realized that infinite series is one of the things I always hated, and besides, even I kind of don't care about math that much.  Still, nice to see that it was totally bullshit, although the Riemann zeta function proof in the link in the comments is kind of interesting.  I did really like complex analysis.
Funny story about complex analysis:  I took that class spring semester senior year, as I needed a filler math class and PDE didn't fit into my schedule (I think.  I don't think I ever took that).  I'd taken a grad level math methods in physics that fall, and a large chunk of that was complex analysis.  Anyway, I get to class expecting that it'll be a more in-depth look at things, probably spending more time doing proofs and shit like that.  Because, hey, it's a 400 level math class, it's probably full of math majors.

It turns out that a lot of math majors are actually math-education majors.  Like "I'm going to teach high school math" instead of "I'm going to get paid very poorly to sit and work out esoteric proofs."

So the very first day of class, we have to go over complex numbers.  Entirely.  How to add them.  How to subtract them.  How to multiply and divide so you again get a standard complex number of form z = x + i y and not some z = (a + i b)/(c + i d).  A whole hour spent explaining this.  There was a EE major in that class too, and other than the "but we use j" wrongness, she seemed to have the same thought of "wtf is wrong with these people."  The professor was fairly young, and I have the suspicion that that first week of class destroyed him.

About the second or third week, I decided that I'd had enough, and went to his office hours to tell him that I was going to stay in the class formally, but I was unlikely to be attending any more lecture sessions.  I show up, expecting it'll be like a quick meeting, only to see THE ENTIRE FUCKING CLASS standing in the hallway waiting.  Apparently they'd been going to all of the office hours, and sitting and doing their homework with him.  We all sit in a circle in this tiny little office, and he's going around and having each person ask a question at a time about the homework.  When it got to my turn, I didn't have a question, so I had to kind of awkwardly say (in front of the ENTIRE FUCKING CLASS), "Oh, my homework is done," [give it to him] "I just wanted to let you know that I wasn't going to be coming to class unless we're having a quiz because we're going super slow, and I already know most of the stuff we've been doing from other classes."  Boom.  Left the office hours.

I know I got a 4.0 in that class, and I really hope that he didn't use outlier rejection when he made the curve for that class.

Thursday, January 16, 2014

Thursday: The rainbows are back



  • Wait...bears aren't musical.  No, this doesn't make any sense at all.
  • Well shit.  That's a big spoiler.  I guess I need to catch up on cool popular shows faster than I do.
  • This is exactly why I love Jigglypuff.
  • Fourier expansion.
  • You can probably do better if you plan your observations better.  Dithering can give you the ability to come up with a flat correction (since stars probably don't change brightness if you rotate the camera).
  • Grant Morrison is a hack.  When he does end up writing something that would be cool, he says that that isn't what he intended, and instead of being a story of redemption, he wanted a guy to be an alien god.  Because that's totally a better story, you fucking hack.

Wednesday, January 15, 2014

Wednesday: I am irrationally concerned with good statistics

First though, there's this Rick and Morty show.  Rick's a crazy scientist, and Morty is his grandson who ends up getting dragged on impossible adventures.  Great stuff.  This week's episode they were abducted by aliens and put into a simulation world for reasons.  However, it's not a good simulation, and Rick points out obvious problems:

Like the guy putting a bun between two hot dogs, and the old lady walking her cat.
 Morty's all "maybe people are just weird, and that old lady is straight up proper crazy."  So Rick has to pull out the big guns:
Morty acknowledges that poptarts probably wouldn't live in a toaster, as that doesn't make much sense.

Nor would they need to get in their car and go to work.

Rick points out the actual flaw in things: people don't drive around in cars that look like their house, but are smaller and have wheels.  Cars and houses are not the same shape, so this is just complete nonsense.
 Yes, they had to be naked there, due to weird alien reasons.
The poptarts show up later, as well.

Ok, statistics again.  The problem is that I saw this article today, which basically complains that "no one really means to use standard deviation, as people intrinsically want to use the mean absolute deviation" which is, of course, complete bullshit.

First, no one would ever do mean absolute deviation in their head.  Here are some numbers: {-1 2 3 -5 1 400}.  If you had to guess another number that would belong to this set, you're going to guess like "dunno, zero maybe?"  You know that 400 is probably bullshit, so you cut it out.  People don't do real means when they filter data.  It's some combination of a mode and median.  Choose a number that doesn't seem crazy.

Second, this mean absolute deviation tells you about where the 50% point falls.  Why that point?  The standard deviation is more inclusive, as it tells you that most (Q(1) = 68.change%) samples are closer to the central value.

Third, all that obvious shit about moments analysis.

Anyway, time for plots.  These are the same idea as the ones from the previous post, just remade with more samples and different stats.  The horizontal lines are the true uncontaminated distribution sigma and the true fully contaminated sigma (sigma_uniform = sqrt((b - a)^2 / 12), because math).  First thing to note:  Actual sigma cleanly switches from the two extremes, as it really should.  Gaussian fits are best, but IQD and MAD are comparable up to the 50% contamination point.  MeanAD doesn't seem particularly good.  The full contamination end is biased, as I'm using a parametric model (that it's a Gaussian distribution).
Biased samples.  This nicely shows that IQD fails before MAD, and that Gaussian fits are reasonable up to 60% contamination.  MeanAD is again off kind of doing its own thing.  Median >>> mean for outlier rejection.


Tuesday, January 14, 2014

Tuesday: I have 217 RSS items that I'm ignoring until tomorrow

Does anyone else think that taking crinkled foil and using your finger to smoosh it back to being flat on your desk is way more fun than it probably should be?

And that's what happens when I don't bother to filter the links before writing the blog post.

Monday, January 13, 2014

Monday: I guess it's going to rain this evening

It seems like a safe bet.

  • Serious stuff.
    • This is why my google hangout stuff doesn't work.  I guess I should try one of these solutions, since google is apparently incapable of making sure "thing they make A" work with "thing they make B."  You know, kind of like google+ and everything else.
    • Did you know that the Kresge behind such things as the Kresge art museum is the same K as in Kmart?  Also, they totally just saved the DIA, so that's good too.
    • Jobs.
    • I don't get Weather Channel.  Probably because a place where it's always the same doesn't need it.  That might also explain the suprise rain storm.
  • Not serious stuff.

Sunday, January 12, 2014

Sunday: Because I'm a grown up, so fuck you

I spent a lot of the afternoon trying to figure out what I wanted for lunch.  Or dinner, since it quickly became 4pm, and you can't call that lunch anymore.  Finally, I saw an ad for IHOP, and that solved my problem.  I'd go to Eggs and Things, as that's easier to get to than IHOP, and sells local pancakes and not industrial corporate pancakes.  

Salad panorama.  The salad makes everything healthy.
Even calimari steak and chocolate chip pancakes.
Other than that, I've spent the day clearing out things on the DVR.  It looks like I'm down to less than 13 days.  I probably should delete entire shows that I have very little likelihood of ever watching 103 episodes of.  I'm looking at you, Buffy the Vampire Slayer.  You're like a quarter of everything I've recorded, and I probably only want to flip through the episodes I liked to see favorite parts.






Saturday, January 11, 2014

Saturday: Excitement on the way to sushi

I noticed a security guy walking all purposefully while heading to sushi.  After putting my name on the list, I discovered that there actually was something going on:
I guess those guys tried to steal from the ThinkerToys?
 Yep.  Thinker Toys.  The place that sells a bunch of toys that don't really look like they're much fun, but they also have some board games.  Well, did have some board games.  They seem to have replaced them with goth jigsaw puzzles.  Who tries to shoplift there?
Those guys, I guess.  Looks bad for them, since the real cops showed up.
 Then I had sushi.
Om nom nom.
 On the way out, I noticed the Sanrio store, and thought I'd take a picture of what it's like.
Kitty.  Kitty everywhere.

Friday, January 10, 2014

Friday: Fish sandwich is about this early, and twitter works right on tv because I see how the lazy weekend is bizarre conniving prostitutes prostitutes

-Or-
So, let's talk about Markov Chains, I guess.

This news story popped up in the RSS today, which reminded me of the long ago time in which Forum 2000 was a thing (note: Forum 2000 never worked like that.  It was all basically a mechanical turk).

I then remembered that you can download all your tweets from twitter, so blammo, I have the text corpus necessary to hack up something to do this.

So, those Markov Chains.  Here's the simple way to think of it:  starting from initial point A, there's a probability p_B that you'll move to point B, p_C that you'll move to C, etc.  However, there are also probabilities for all those points too, so you can chain things together.  Maybe go look at the wikipedia figure.  This isn't turning out to be that simple of an explanation.

In any case, if you let each "point" be a word in a tweet, then if you have a large sample of tweets from someone, you could imagine that you could construct fake tweets if you knew the probability that a given word follows another.  That's what I did in perl, and that's where the title came from.  I did a simple chain, where w_{i+1} is drawn directly from the probability distribution of words that follow word w_i.  There's a reset condition where if the set of words that follow w_i is empty, I restart the chain from the set of "words that start a tweet".  I've also forced the length of the generated string to be 140 characters, the standard tweet length (and forced everything lowercase, and removed all punctuation, etc, etc, etc).

Here are some more examples:
  • i was far the whole dinner but that @wholefoods sushi was aborted because its weird being even fucking clue what hes a cheater :-p theres no
  • hey verizon network that too hungry or more not sure that flasks were afraid the sun orbit simulator its now its someplace else too soggy as
  • happy birthday @jkru isnt helping anybody know and hes never a 12 16 garbage bags tied directly to work - british accent for being wasted its
  • rt @vgc_scott man that quote and cranky tonights dinner oh man its still preparing the oil is actively looking up a workaround to be going to
  • first time you dont want a walk an impenetrable island which is like it looks close eyes will always just woke up dry before i not having intermotrons
  • @jkru its all the music #fuckedupdream peter potomus cartoons are you your app its official word for liberty tsa approved for my implementation
  • dear this is what is terrible monsters outside who like fog and colby isnt even left there to the 100th day yay deposed monarchy ok so hopefully
Those are just nearly not gibberish enough for you to think that they're not computer generated.  The crazy thing about MCs is that they do a wonderful job of constructing sentences that are nearly grammatically correct.  This could probably be improved a lot, such as trying to pick words that have some influence from the second previous word or having it clearly mark where it's restarted the chain with a period or something.  Still, for like ten minutes of work, this isn't too bad.

Finally, an interesting sampling from printing out the chain transition probabilities:
fuck you 0.294392523364486 63 214
fuck it 0.0794392523364486 17 214
fuck fuck 0.0700934579439252 15 214
fuck yeah 0.0700934579439252 15 214
fuck up 0.0327102803738318 7 214
[...]
fuck #sandwichtweets 0.00467289719626168 1 214
fuck shitshitshitshitshitshitshitshitshitshitshitfuckshitshitfuck 0.00467289719626168 1 214
[...]
fucking ghost 0.00454545454545455 1 220
[...]
fuckity fuck 1 3 3

This isn't really related to anything, but it kind of looks cool, so that's something.