meta-roj

arts and urology

evolution

actor

land

in a sort of follow-up to this, we have another interesting police-tactics video to consider.

Video: Police Handcuffed 5-Year-Old Girl [ap via abc news, april 23, 2005]

A 5-year-old girl was handcuffed by police after she tore papers off a bulletin board and punched an assistant principal in kindergarten class, according to a video released by a lawyer for the child’s mother.

The 30-minute tape shows the child appearing to calm down before three officers pinned her arms behind her back and put on handcuffs as she screamed, “No!”

i haven’t seen the video, but you have to wonder how much of a threat a kindergardener really poses these days – they probably don’t even have scissors in the classrooms anymore or they wouldn’t be able to get insurance.

this is just a quick one, from wisconsin…

Hopping Holstein shuts down cow pie bingo [lacrosse tribune, april 21, 2005]

Dorothy Jasperson, editor of the Westby Times, said the cow “knocked down some stuff, scared some little kids and for insurance reasons, they decided to go away from cow pie bingo.” (There was a time when Westby children were not afraid of a wandering cow, though these may have been children visiting from La Crosse.)

no insurance, no play.

wow. it really has come to this?

there’s trouble in river city, or, at least, in santa cruz.

the cameras are everywhere. everything is on the record. let the spin begin.

start here, and see if you’re comfortable in the new america.

[via boingboing]

i’m not a comic kinda guy, but i am a culture kinda guy. so a comic of a different culture is worth a few words. now that they’ve been around for a little while, i’ll give them a pointer: ak comics.

back at the end of january/beginning of february, i stumbled into two pieces on how the music business was flopping around trying to find some new models to survive on. this was about music video distribution.

i don’t want to get too far into the history of the music video, but i will say that generally speaking, the music video has been viewed as an advertisement for the album, much like the tour was to “support the album.” the album, of course, being the unit that matters for the record company, since their revenue structure is, historically at least, built on album sales. the current trouble probably has something to do with the death of the album.

in any case, it’s obvious that the record companies need to find new revenue models, and they’re willing to try anything.

universal and warner have realized that the music video is a product unto itself, and as such can be sold (or, rather, monetized):

Universal Music Charging Video Nets to Air Artists [ny post, february 1, 2005 – now locked up behind a pay-per-view model itself, and certainly not worth $3.95]

Universal’s new policy is believed to be the first in which a record company has decided to charge per video for every service that exists.

MSN is the first to sign on under Universal’s new policy. Universal’s existing deals will remain in place and be renegotiated later.

then there’s warner…

Warner Music Group and Verizon Wireless Launch America’s First Music Video Download Service [investors.com, jan 31, 2005]

Unlimited basic video clips are included, however, premium content such as music videos and 3D games are available for an additional fee, including WMG music videos for $3.99 per music video.

many, many years ago, i spent a good deal of my life trying to solve some really brutal combinatorial problems with every widget i could find in the geek arsenal – i hacked on trees and heuristics and neural networks and genetic algorithms and all kinds of strange combinations thereof to coax optimizations out of personal computers.

this was an affort to improve on human optimization of really large combinatorial problems, and do it interactively, so the “human intuition” could augment “brute force” (and vice-versa) to push the solution closer to a true optimization rather than letting it get stuck in a “local optimization.”

during this time, i ran into the amd 286zx and 286lx processors (i think that’s right), which were essentially systems-on-a-chip. these are now so ancient that they don’t even show up on amd’s site anymore. anyway i had some crazy idea about packing a whole stack of these little 10mhz parts into a desktop-sized box with a 386 playing arbiter and doing user-interface duty. i figured that asymetric multiprocessing would be an interesting approach to the problem, where some small piece of the problem was assigned to one of the “slave cpu’s” which would be allowed to grind away on it until the “master cpu” (perhaps prompted by the user) decided that it was spinning its wheels, killed it and gave it a new subproblem to chew on for a while. i figured i’d start with the 286zx’s since they were available, but could slip in any sort of optimized slave processor as it came along (say, chunk of silicon design specifically to run neural network simulations or whatever). it was fun to think about.

ultimately, though, i ran out of money and ideas and patience and never quite got to a functional solution (and couldn’t afford to build custom hardware).

i bring this up now, because it sounded errily familiar when i stumbled into this news release from cornell.

Mostly their approach is to have the computer do what a human being might do: stop, go back and start over and try something different.

these people are definitely smarter than i am, but it’s neat to see it crop up again.

one tip for the cornell team: it’s possible to get some really, really neat visualizations out of these problems, and if you tweak them a bit you can get some wicked user-interaction experiences.

i bet current graphics processors would be really good at this stuff – both the visualization and the actual problem-solving. a lot better than the 286zx’s, anyway…