Sunday
Feb172013

Jonathan Coulton ripped by Glee?

I rather like a couple of JC's songs.  Can't count myself a loyal fan, but it comes as a disappointment to me that he's had trouble with FOX.  After getting the rights to "Baby Got Back" by Sir Mix a Lot, he created an seemingly entirely new arrangement and melody that sounds tender, is about awkward self realizations and is humerously lovestruck.  The original's style seems squarely aimed at loud obnoxious objectification with a good humpy beat. 

His problem comes from FOX's Glee show having purchased the rights to "Baby Got Back" from Sir Mix a Lot's label, and then having used JC's novel arrangement without any compensation or credit.  His legal recourse seems limited.

So what is a man like JC to do?   Apparently a twitter campaign.

What's done is done, but a musician might avoid future similar situations by releasing a wholly self-owned instrumental version before releasing a lyrics-only cover.  This might bifurcate the ownership rights a bit more clearly.  

Karaoke fun, anyone?

Wednesday
Jan302013

Aquatic laser array - satellite based mobile listening outposts

Interesting link here:  http://www.atimes.com/atimes/China/LE13Ad01.html

it appears that a laser distance scanner mounted on a satilite can scan a block of ocean, report acurate ocean wavelet heights and work out the locations of every big fish in the sea... or act as a point-it-anywhere submarine hydrophone.

Pretty easy for me to see how such a device could make a mobile listening outpost... it's trivial to beam a laser it at a window and route reflected antenna (sensor) signal to a mic.  I could do that much back in high school.  But building a underwater radar picture from a wavelet map seems... well at the very least tedious.  Maybe a deconvolution matrix mixed with a fourier transform would make it possible?  Probably would want long wavelength - far infrared - so that it could zip through clouds.  

 Random daily thoughts.  Maybe I can get one of my kids to work on this idea for a science fair.  Proof of concept with optical laser would probably work.

Tuesday
Jan292013

Pythagorean addition - why doesn't everyone use this?

Nonsensebox (or Plausible or whatever I'm calling the little beast these days) is a almost-not-quite artificial intelligence that looks at large data sets and automatically makes conclusions.  Many many conclusions, linear or wildly nonlinear, logic rules or purely numeric operations.  It lets me look at a dataset and search for meaning provided that dataset is healthy and big enough the the degrees of freedom stay tame.

And Nonsensebox generates lots of coincidental nonsense trivial.

I use Nonsensebox for numeric multivariate analysis - and perhaps not that unexpectedly it keeps coming up with equations and relationships any scientist or financial planner might be intimately familiar with.  Banal stuff.  But it does other things too.  I've noticed that it keeps reporting relationships in multivariate spaces as variations on Variable C=constant*(constant*Variable A^2+constant*Variable B^2)^0.5.  That's the pythagorean theorem!  

So I added pythagorean addition as possible relationship function to Nonsensebox.  

And in multivariate rule determination, Nonsensebox uses ⊕, Over and over again.  (The symbol for pythagorean addition is ⊕ in specific math dialects, but apparently it means other things as well like Earth.)   So I wonder, why is ⊕ something covered like a oddball little bit of math in grade school and for a couple hours in high school and college when really it looks like fundamentally, ⊕ is as central to physics as say the division symbol?   

Try throwing ⊕ into the relativity equations and you get an simple equation that includes something that looks a lot like compelling explanation of inertia.  Anyway, if you see evidence of pythagorean addition in your equations, I recommend you rearrange with the ⊕ symbol instead.  After all, C=A⊕B is so much more beautiful than (C^2)^0.5=(A^2+B^2)^0.5. 

There is just a chance that messy modern work-arounds and ugly notation make mathematical concepts with this elegant function less clear and hardly obvious. 

Thursday
Jan172013

My child, future truck driver (A rant: What the hell is ACT-Explore about?)

My eighth grade son H brought home his Illinois ACT Explore Score Report today.  It intends to counsel plans for education, career and work after high school, and in our school district, test scores influence curriculum choices in High School.   

In full disclosure of pride before a fall, I'll report that H tested above 99% of other students taking the test in the United States.  His score in mathematics was above 100% of the students in the united states (which should bring to mind rounding error).   His lowest score was in English, still better than 89 percent of his peers.

What career counseling does ACT offer my son on their World of Work Map?   He is apparently unfit for careers in management, marketing and sales, politics, education, communications, health care, law, the arts, the financial sector, medical technologies.   Instead he should consider Truck/Bus/Taxi driver,  Construction and Maintenance,  Dry cleaner,  Agriculture or forestry, Auto mechanic.

Looking at the "Understanding your EXPLORE results" booklet showed an example score at the 69th percentile in English,  the 36th percentile in mathematics and 15th percentile in Science and suggested they might aim for careers that include management, law, health care, education.  

So, this made me think, what do the ACT EXPLORE results mean?  Is the councel it offers for the benefit of the child, the benefit of society, the benefit of the school counselor, or for the benefit of former Hawkeye Iowa City ACT child development hipster psychologists that may (may?) have goodwill but really don’t have a f-ing clue and mostly wish to sell a new products to one governing bureaucracy or another.  

So, to my son H:  Your ACT Explore results are busted, bogus smoke being blown right up your ass.   It won't be the first or last time someone tries to do that.  Don't let that hurt you, informed opinion isn't knowledge.   If you want to know why your results are that way it's probably because you are an edge case their models don't understand and because you answered preference survey questions in a guarded way.  So please H, Do what-ever you want with your life.  Unlike some of your peers that haven't worked as long or as hard to get perfect grades and haven't earned your skills and talents- I'm pretty sure you can do anything you want with your life.  Nothing is beyond your grasp.  No career is unsuited to you - should bus driver or dry cleaner inspire your passions, go for it.  But please find that passion and make it happen.  I've seen what you can do.

Thursday
Jan172013

Mathematica 9

The new version of Mathematica is better - almost fun.  I'm not certain I can chalk it up just the HoeffdingD function.   Lovely function that one tho, actually fantastic, and I'm glad to have it*.   Perhaps I've mellowed, I enjoy Mathematica a lot more than I did a just a while ago.  I feel more like I'm collaborating with the program and less like I fight for it do my bidding.

Thinking about it a bit, my number one feature of Mathematica 9 that increases happiness must be the little spelling-function-variable reminder window - a thing that as a version 8 user three months ago I would of told you was worth $0 to me.  In fact, I'm getting older, and there's that long healed ski concussion to consider.  My old step-1 workflow of "memorize everything" just isn't low effort anymore - So when I wonder, "did I call that thing rowselectiongenepopulation or something less verbose?" I only have to type a few letters / cursor down and enter.  This little change makes it effortless to stay with the thought flow and avoid my natural misspellifications.    

 Syntax error highlighting in the notebook is improved, but not quite praise worthy.  It is much easier to see brace, bracket and parenthesis given messy-pile-of-shoelaces nesting, but there is not quite enough highlighting for Function[SubFunction[List[[item3]][[elements2;;60]] ]]; to have unique highlights for each brace combo or type of semicolon use.  I won't claim to be a normal mathematica user- I almost never use anything in the "Palettes" assistance menu or Format pull down menu so there may well be great ways to customize auto-syntax highlighting and make it more like an "integrated developer environment".  

The second serious improvement in Mathematica 9 is that Mathematica has gotten better about telling you what it can do.  At first glance the "more functions/common commands" palette hardly seems like more than a toy and typically doesn't prove me with a forward workflow - BUT - it is good at introducing code functions or concepts that otherwise would of been scattered in the ?four thousand++ pages of online help.  And when used occasionally as an alternative to the online help, results can be brilliant. 

I work with largish data sets and Mathematica is awkward.  Branched nested data structures, polling data, creating tables, selecting specifics-- its still a chore to code so as to keep memory use and CPU load minimized.  All the same, my problems with data and perfectly reasonable looking code that runs 420 times slower than obscure arcane methods probably are not typical critisims of Mathematica.  Still, I wish Mathematica naturally tended toward fast and lean with regard to datasets.

All together, Mathematica 9 almost introduces a feel of play.  And I mean 'play' as extraordinary praise.  My only disappointment is that my own children (about 11 and 14 year old math wizards, but not computer wizards) don't connect to the joy.  Dad's just a little out there when it comes to crunching numbers and searching for unusual statistical relationships.   I hope that future versions of Mathematica will continue to expand on the concept of playful discovery and accessability - this particular dad wants his kids to discover the same sense of wonder that I found back in ?88 when mom was building Cray supercomputers and a nifty ?NSA? analyst affiliated somehow with Cray introduced me to Mathematica 1.0.   Or maybe my kids just don't get it because there are no spy agencies or super computers in their humdrum lives... hmm... could work on that I supose.  Seems like that would be complicated tho.  Hmm... better not.

 

 

 *The new function, HoeffdingD[list1,list2] seems to take issue when comparing a target with a list that has near underflow differences between its values, and this unusual situation may sometimes result in reported correlations outside of the expected range of -0.5 to 1.0 for this function (I have found results greater than  1.4), but all this messyness is avoided with Round[lists, 0.0000001].  Or Round[a few less zeros than machine precision]