Friday
Apr112014

Mathematica 10? Wolfram Language

I've gotta say that I'm terribly interested in Wolfram Language - although I suspect it will remain a bit more web-centric than my own tastes in software run.   What makes mathematica so amazing is that if you can conceptualize the essence of a problem properly in the language, a hundred lines of code can often do amazing things that some other computer languages might accomplish in ten thousand lines of code.   Of course it's not such an easy feat to "conceptualize" and describe problems - and its possible to implement a search for solutions on a model that has no validity in our universe. 

 
But those are problems in thinking about problems.
 
I'm going all "six-year-old kid in a candy store" for Wolfram Language; just can't wait to get my grubby fists sunk deep in those bins.  Just wondering if my ideas will still be my ideas if the computation is completed by a corporate cloud.  Never-mind. O-boy!  Oboy oboy oboy!   :)
 
Friday
Jan172014

Lamentations on Steven Jay Gould

 

I visited my local library this week.  While perusing the stacks I came across Steven Jay Gould's book "The Structure of Evolutionary Theory" and found the hours slip by.   For a blogger that didn't read even a tenth of the ~1200? pages I'm going to offer stiff criticism - it reads like a fussy historian's blog.  

I didn't spot the data or mathematical models I might of been looking for.  More eloquent histories and pictures of Cathedrals than apparent grounding in mathematical theory.  

Steven Jay Gould was a childhood hero of mine.  His insights into biology were extraordinary.  I miss reading his regular columns, and he died too soon at 60 years.  The Structure of Evolutionary Theory is a terrible stand-in for Mr. Gould himself.

Words directed at the next biologist superstar (a species more rare, apparently, than superstar physicists):  Consider great collaborations with a brilliant mathematician.  The division between evolution and information theory is just about specifics. 

It seems perfectly obvious to me that genetic algorithms can find workable solutions to problems well beyond the grasp of a bright human mind.  The evidence that they can and do is nearly everywhere.

-D

 

Thursday
Jan092014

iCitizen - a neat concept with skewed data

iCitizen  is a great app for iPhone that gives you a 360° view of local regional and national politics.   You can select to follow issues, elected representatives, get voting records.  For the first ten minutes of playing with the program, I was really impressed.  But unfortunately, a rather skewed peculiar view of politics quickly emerged when the program started pushing me to answer polls it conducted.

What would the world look like if you only asked Lamborghini owners what it should be like?    I think it's fair to say that you would have unconventional views concerning the next presidential election, our elected officials, and what issues are most important compared to average voters.  

iCitizen does something similar with expensive phones.  "Should undocumented immigrants be granted amnesty?  63 percent of Nebraskan's think so!  But actually just twenty Nebraskan's voted, wise or  foolish, the Nebraskan iPhone owners that use iCitizen and respond to polls may not represent the body or will of the people.  

Heck, with a VPN that exited in Nebraska, I might be able to swing the vote more than a few percent just by pooling my votes across my ios devices.  Maybe that explains some of the results I've seen- hackers that want Alaska stuuupid.  

So iCitizen, nice idea - like that I can see how many votes my elected officials have skipped because they apparently had better things to do than represent the electorate.  I don't like that iCitizen misrepresents the desires of iphone/android users intrested in politics that purchased this specific application as America.  The notion that this skewed demographic is somehow America is a bit frightening.

So, fun application - but more informative about your demographic than about your world.  And the notion that an elected official would use poll information to inform their decision on one of those rare occasions when they actually found the time to show up for a vote is unsettling.

Tuesday
Jun112013

Paranoid leadership unjustified?

The NSA (Nationwide survalence agency?) Prism scandel has been breaking for the last few weeks and I have two cents of opinion to share.

First the upsold notion that these programs are worthwhile because they have prevented terrorism seems at odds with the cost of the programs.  Of course, the exact cost is unknown to me.  Today's wikipedia entry describes it as a $20million per year cost, but given that a $6billion dollar/year company exists solely to fullfill 2/3rds of the program needs, it seems likely to be close to a 9 billion per year set of government contracts.

For perspective, that's cost equivalent to roughly five Moore-OK F5 tornados worth of terrorist causalties per year.  Are we preventing that order of damage from terrorism on a yearly basis?  I think the answer is no.  Boston... Newton.... Santa Monica...  Army Staff Sergeant Robert Bales premeditated murder of 16 Afghan women and children, the bad stuff still happens.   Prism doesn't amount to cheap insurance at all.

A second question, is it worth it for the US government to intrude upon us in this way?  Yes.  Its quite likely that nearly every government is trying to squeeze metadata from credit cards, phone records, email, skype, facebook, even if they have to write spyware to do it - very much similar to the situation advertisers do.  It is in our national intrest to have intelligence.   

I just wish they were open about it, perhaps because I niavely think that sunlight would prevent the degeneration towards corruption and abuse.  If FISA Courts were once a protector and are now a rubberstamper, would that still happened in the daylight?  

Friday
May102013

Evolution - That thing that everyone thinks they understand far better than Darwin.

Biology types:  In the United States, you are loosing mindshare to lax-thinking young earth creationists.  The notion that evolution isn't real needs to be challenged.  And frankly sticking to the beautiful but frequently impenetrable ideas laid out in a book published late in 1859 seems not to be helping.  

Simply, the idea of evolution needs to expand past Darwinism - consider an information theory based definition of Evolution - Evolution can be considered a circumstance that results in successive refinements to a design.   Yep, I used that word "design"... don't flinch, own it 100%.  American pronghorn antelope populations were designed by the the grass they ate, by hunting wolves and extinct american cheetahs, by dry seasons, competition with other herbivores, competition among males, female antelopes selecting for "desirable" mates, and perhaps even wild fires and volcanos all in the quest to reproduce again and again.  Lean into that awkward young earth creationist idea of a designer and own it - tell the story of who the designers were and how fat short legged antelopes don't exist because the cheetahs ate well and the females refused and show that evolution lets even small antelope brains direct design choices. 

Mathematical types:  Please, pick up the slack.   Evolution is an astounding example of game theory, successive iterations of reward or failure that drives changes in competitive populations.   I use evolutionary algorithms to provide practical solutions to deviously tedious problems.  (Sure not exact solutions, but practical.  Almost always better than engineering's "simplify by considering this turkey to be a perfect sphere" type approximations.)  It pains me that a rather complete search brought forth not quite 200 titles devoted to the mathematics of evolution.  I have very specific quandaries about algorithm strategy optimization and computability and the sum of all present published knowledge on the subject is... not all that helpful.  There isn't a hard wall between mathematics and biology, study evolution and the ways it bridges and marries these subjects.  

Computer types:  Evolution is an incredible example of complex information theory.  DNA is code.  Nearly all life runs in wet jiggly ~6 bit computers.  And sex makes the configuration complexity staggering.   Nearly every human, fish, orchid, sponge, spider or worm is running a unique set of code or code configuration.     Diversity happens automatically with each copy attempt, there is just so much to copy that a change or two will creep into the spools of code, and those little glitches usually get weeded out if they interfere much with reproduction.  Leave the code running in one species for too long and you'll get so much diversity in configuration and code that anyone would swear you have a thousand species of distinct creatures, almost like you started with a vacuum tube mainframe and a few thousand competitive iterations later had an Ipad, Google's server farm and Facebook.  

So what has me on this kick today?  I watched a documentary, and it made me feel like the United States is sprinting toward stupidity for election day gains with the young earth creationists.  I'm all for artistic, religious or political freedom- but even I can't stomach the cost of dumbing down a generation or more of our children.  I care about our future.  I don't think that being taught well intentioned dreams over well known principles of biology, mathematics or information theory has decent crack at improving our world or advancing the United States position within it. 

Evolution isn't just a theory, its a process, a situation, an activity.  Embrace it.