Wednesday, September 01, 2010

The Cray 1A resurrected

For people of a certain age, Cray was the ultimate mind blowing supercomputer. Several of my colleagues worked for the company.

So how fast was the original Cray 1A, the monster machine purchased by top secret agencies to build bombs and do spy stuff?
Homebrew Cray-1A | ChrisFenton.com

... Now, let’s get down to specs - What is this bad boy running? The original machine ran at a blistering 80 MHz, and could use from 256-4096 kilowords (32 megabytes!) of memory. It has 12 independent, fully-pipelined execution units, and with the help of clever programming, can peak at 3 floating-point operations per cycle. Here’s a diagram of the overall architecture...
My iPhone 4 is ten times faster.

We really don't understand our world.

As you might guess from the phrase "What is this bad boy running?" Chris Fenton is an absolute, stark raving mad, uber-geek. He has built a Cray-1A equivalent at 1/10 scale. The only thing he's missing is the software. Maybe some of my friends have some at home.

Read the story ....

Cable connections - now think about conflict

Fallows blogs (again) about the infowarrior global telecom map, which includes cables like C2C (emphases mine):
Greg's Cable MapC2C

7.5Tbps

16 Cable Landings  
Changi (Singapore)
Nasugbu (Philippines)
Chung Hom Kok (Hong Kong)
Tanguisson (Guam)
Chikura (Japan)
Redondo Beach (USA)
Hawaii (USA)
Los Angeles (USA)
Hillsboro (USA)
Nedonna Beach (USA)
Toyohashi (Japan)
Shima (Japan)
Pusan (South Korea)
Nanhui District (China)
Tamusi (Taiwan)
Fangshan (Taiwan)
War would be very inconvenient in the modern world. James F sends us to Stephenson's classic Wired article on this topic, I learned some of that history from Stephenson's later Cryptonomicon.


Tuesday, August 31, 2010

Pawlenty - an opportunist to despise

Tim Pawlenty is governor of Minnesota, with about four months left to serve. He's not running for reelection. He's supposedly running for the 2012 Presidential election.

Given his time remaining, this announcement will probably have limited impact. It's a good reminder, however, of what a sleazebag he is ...
Pawlenty restricts health money | StarTribune.com


In a move that could cost the state $1 billion or more in federal health care funds, Gov. Tim Pawlenty announced an executive order Tuesday designed to keep what he terms "Obamacare" out of Minnesota.

Pawlenty said he will require all state agencies to funnel their federal grant requests through his office in order to "stop Minnesota's participation in projects that are laying the groundwork for a federally controlled health care system" -- unless they are required by law or approved by his office...

... Pawlenty could be closing the door, at least during the remaining months of his term, to more than 100 federal health care grants that would fund projects ranging from diabetes prevention to postpartum care for new mothers to tighter regulation of insurance companies...
Think of Pawlenty as a brighter and more cynical version of Michelle Bachman.

Cross-Country Skiing in MSP - Adelsman

The web is a lot like LA. It has no history.

Almost no history. By web standards Adelsman's XC ski pages are ancient civilization. I linked to them in my 1995 skijoring page (long neglected) but they're still around, and better than ever.

I found them again when I wondered where a marvelous paper listing of Twin Cities trails had come from. Google showed me it was a reformatting of Adelsman's metro ski page.

They do take advertising, but, as marvelous as this site is covering Michigan, Minnesota, Wisconsin and Iowa, I doubt they get rich from it. It must be a labor of love.

Our MN state fair visit is past, the days are getting shorter, and now we pray for a rerun of the excellent XC ski season of winter 2010. We'll be using this site, including the lodging section (hint to advertisers :-).

PS. Naturally skinny skiers are big on tradition, but a feed would not be amiss...

Gmail's personalized classifier: your own private AI

When you read about Gmail's new "Priority Inbox" feature, it helps to know that "Classifiers" are at the heart of what we used to call "AI" ...
Gmail Priority Inbox
....Google has to build a personalized classifier for each Gmail user and it needs a lot of messages. 'Email importance ranking works best for people who receive a lot of email,' explains Google. Google takes into account implicit signals like: the messages from people you frequently email are important, if a message includes words frequently used in other messages you usually read then it's probably important, the messages you star are probably more important than the messages you archive without opening. There are also explicit signals: click on the important/unimportant buttons, create filters to mark messages as important."...
Classifiers are used in speech recognition, Google Search and so on -- they're not new. For that matter, this kind of "what's important" ranking has been done many times without using formal classifiers.

Still, this is a mini-milestone of sorts. Classic AI technology is moving to our personal lives from multiple directions. The next step is deploying these classifiers into Google's Facebook-alternative, and then it's another step to our pending lifestream classifiers (which might help keep memory-impaired boomers functional longer).

Yes, those are Skynet's footsteps you hear ... :-).

Yes, this is part of why the IT driver of our whitewater world is not going away, but only getting bigger.

Sunday, August 29, 2010

Michael Lewis - The Big Short - how we lost our economy

Tim Bray has written a review of Michael Lewis book "The Big Short". Tim loves Lewis book.

Do read Bray's review and lessons learned. Lesson 1: People in the Finance industry are in the business of making money for themselves. Obvious, but easily forgotten.

Bray gives us some links to read. Here they are, followed by some of mine:

Tim's links:
Old posts of mine:
It all feels like ancient history, but it was only two years ago, it's not fixed (especially the rating agencies from hell), and most of us are still unsure we really understand what hit us.

Of course I think it was fundamentally computers and China. Robotics, AI and Peak Oil lie ahead, so get used to the whitewater.

Answering Krugman: Why the Obama administration isn't panicking about the elections

Paul Krugman doesn't understand why the Obama administration isn't panicked about a pending electoral catastrophe.

I think he believes they're delusional.

That could be. Denial is a prerequisite for the happiness of a sentient organism. We're good at denial. We're also prone to assume that our local slice of humanity is representative of the nation.

So it would be normal primate behavior for Team Obama to expect catastrophe will be averted.

Except -- they're smart, they're professional politicians, and they don't hate science. So I don't think they're delusional. I think they know what's going to happen this November.

They know the GOP will take at least one of the House and Senate, and probably both. They know this will be very bad, but they expect the Senate will not be able to override a filibuster or a veto. (I will applaud if the GOP ends up trashing the filibuster, though the consequences will be short term terrible.)

They also know that this outcome was decided when the the economy failed to recover. It doesn't matter what they say or do; American elections are decided by "undecideds" and "independents", and those voters typically have no connection to reality [1]. They respond to personal circumstance and mass feeling -- and there's no way either of those are going to change as the Great Recession grinds on.

So Team Obama isn't panicking because that would be pointless. We have already entered a state of political paralysis, and this will probably persist from 2010 to 2012. There will be some opportunities for "lame duck" actions and recess appointments -- and that's where the political pros are fighting now.

Will Obama lose in 2012? It's certainly possible. I fear the Great Recession is going to grind on well into the Peak Oil era, in which case we'll start calling it GD II. On the other hand, I never expected him to win either. Don't underestimate Obama.

[1] Really. The political science profiles are definitive.

RV timeshares today and tomorrow

We've got 3 Kids around 11, 1 dog, an old van and a newish Subaru wagon.

A new van is the default purchase, but most of the time I just need a commuter box. For our epic family road trips an RV might be cheaper and better.

So it occurred to me we need a timeshare RV. These days, to think of a business is to discover it. Turns out, RV Timeshares exist.

I don't know how well they work. There are all sorts of perverse incentive/tragedy of the commons problems in the timeshare industry. I expect they've been a fringe industry.

That's likely to change as we near Peak Oil [1] time and as the Trilateral Commission [2] prepares to implement a Carbon Tax. My family doesn't need a van most of the time, and when we do need it a camper/RV combo would be preferable. If gas were more expensive a market would develop to provide RV type vehicles -- and I could buy a commuter car.

We just have to wait.

[1] Yeah, it's coming. We'd see gas over $4/gallon now if not for the fact that the economy of the industrial world cratered. Given the way things are going the Great Recession will slide right into Peak Oil territory, in which case we'll just call it GD II.

[2] I was a member! Or at least I got their mailings in College. Maybe if it'd been less chatty I'd be a member of the Great Council by now. (PS. This is a joke.)

Guided evolution in cancer therapy

Another of my speculation is cheap (but fun) posts.

I searched Google Scholar today with the terms guided evolution cancer. I didn't find anything interesting. I bet, however, that in 10 years a similar search will have many hits. Today, for example, a search on guided evolution HIV gets some tantalizing hits.

The idea is easier to understand in the context of HIV, particularly if you know the evolutionary history of  parasite/prey relationships. So I'll start with predator/prey, then HIV, then Cancer.

When a pathogen, like the rabies virus or smallpox, is introduced to a new species it is often very lethal. Over time it often transitions from parasite to symbiote. This transition involves adaptation by both host and parasite. This happens because one of the best ways to "get ahead" in the living universe is to "get along" -- particularly if your host shows "promise".

HIV (and other retroviruses) have such mutable genomes that they evolve within the lifespan of a human host. Modern HIV therapy takes advantage of that. Therapy tries to trap the virus into an adaptation box, where one adaptation compromises another. I wouldn't be surprised if the newest therapies try to guide HIV to evolve down a path where a yet-to-be-used drug will be particularly potent. That would be guided evolution in an HIV context.

Which brings me to cancer. All physicians know patients who have lived many years with a metastatic cancer. The malignancy is not gone, but neither does it kill. The host and the "parasite" are "getting along. In time the tumor will kill, but of course we all die. The goal of medicine is not (yet) immortality, but rather deferred morbidity.

If we think of a cancer as a competing set of "parasites" (for the cancer is genetically diverse) laying claim upon the resources of the host, then can we guide the evolution of that ecosystem? Could we use one set of drugs to guide the development of a subset of the cancer that will outcompete other tumor lines? That successful subset might be so guided as to be very vulnerable to a "knockout" drug -- or to persist but in a relatively tame state.

Just speculation. I'm only about 50% right with these sorts of things.

Survivors and Foragers: speculation on group selection and obesity

Speculation is cheap.

What if one of the variables in human weight control was activity response to caloric restriction?

Ok, so that's not too speculative. We know obese humans decrease activity when they diet. This is one reason that, in a nation of very cheap calories, so many Americans are obese. When these people diet, they are put into an involuntary lethargic state.

It's a behavior that makes lots of evolutionary sense. Throughout most of human history these people would have been survivors of famine. We can call them The Survivors.

But what if there's a distribution across the population? What if there are people who respond to caloric restriction with increased activity? That makes a sort of Darwinian sense too, particularly if you think of humans as group selected Big Brain Bees (BBBs). Think of these people as Foragers, they roam from the Hive looking for new options. In todays world, some of them are anorexics, but mostly they're just slender.

I am positive that if I knew the scientific literature on obesity, and I knew the right search terms, I would find articles exploring this model. It feels plausible that there's a range of responses to calorie restriction -- from Survivors to Foragers and everything in between.

Of course feelings are the fount of grant proposals, not of science. This is, however, a testable hypothesis. So it could turn into science.

There are implications for drug discovery. If this model were correct, the magic obesity pill would convert Survivors into Foragers. So pharmas should be looking for Foragers to study ...

Friday, August 27, 2010

Google's bicycle directions are getting scary good in MSP

A few weeks ago Google's bike directions from my home to my office were pretty good, but they did send me over a fairly narrow and crowded overpass. Still, I was impressed. It was as good as I could do at the time.

Today I remembered a really obscure hidden footpath that would be an improvement on that route. I went to Google Maps so I could provide feedback to the bicycle map team.

The route improvement was already there:

Wow.

PS. Don't get too cocky Google. Your "draft" Blogger rich text editor still can't handle European language paragraphs. My theory is it's being developed by a team unfamiliar with the paragraph.

Thursday, August 26, 2010

Anthem - Putting the Hell in Health Insurance

My corporate health care insurance is administered by Anthem.

Tonight we tried to find out if a particular physician is in their coverage network.

First I tried their Account Registration. It includes a CAPTCHA test.

Why does it include a CAPTCHA test? Really you don't need to ask. Since CAPTCHA hacking software exceeds human performance, it's really a filter to eliminate humans. Particularly elderly humans who might need expensive services.

I kept failing the test. Emily checked, and I seemed to be typing the correct response. We even tried the audio test. That one would  have impressed Mephistopheles.

So Emily tried their customer service number. Voice recognition of course, no keypad entry. It couldn't understand anything she said. Her voice kept rising, but it didn't help. I tried not to laugh too much.

So I tried the web site again. This time I got a different response: "Sorry, we're experiencing technical difficulties at this time. Please try again later or contact Customer Care. Error ID: 21573301".

Terry Gilliam couldn't have done it better.

It's moments like these that make me more optimistic about the future of humanity. Our civilization is sure to collapse before we can create the sentient machines that will end us. We will be saved by our own greed and incompetence.

PS. The coup de grace came as a post-visit survey. An opportunity to provide feedback! So I started answering questions. The questions kept coming. I noticed the the scroll bar size -- I was only through 15% of what must have been 50 questions. I tried scrolling to the end and submitting. Of course that was rejected; I hadn't answered all the questions. I had to smile at the style of it; sadism with a flourish.

Update 8/27/10: There's an old joke that the best way to improve health insurance plan revenues is to put the registration office on the top floor of a building with no elevators. Only the healthy can join, so that even low rates will be profitable.

The CAPTCHA is the modern equivalent of the top floor office without elevators. In this case there was a web site failure, but I'd wager their web site is always fragile. Since Anthem has already been paid, their incentive is to deny services altogether.

The beauty of these sorts of scams is that they're emergent. They develop in the same fashion as antimicrobial resistance in bacteria, or lousy service in AT&T's flat rate data network. Entropy and funding choices provide the "mutations", the commercial market provides the "selection", the system irresistibly evolves until we get the Anthem web site.

This kind of perverse incentive is built into many systems, but it's particularly strong in the health insurance business. If we recognize and understand these processes we can work around them -- much like modern HIV therapy works with an understanding of the the virus evolves. I suspect oncology is going to go the same way -- using therapies that "guide" the "evolution" of the cancer / tumor ecosystem towards forms that may be most vulnerable to a 2nd wave treatment -- or most compatible with the life of the human "host". (I kind of like that cancer idea btw, I do hope it's really part of modern therapy.)

As a society, however, we're only in the most early stages of understanding the evolution of emergent corporate dysfunction and how to manage it.

Wednesday, August 25, 2010

Who does phone surveys anyway?

I'm a grateful reader of the FiveThirtyEight Blog, now hosted by the NY Times. They give everyone access to electoral data that was once available only the pros.

Of course their data is only as good as the phone survey. I've been wondering, lately, how well that is working.

We get quite a few "survey" calls. Some of them are surely legitimate, but others are ploys to raise money for campaigns. Some are probably "phishing" (identity theft) operations. Given the range of "survey" calls we get, it's not surprising that we never answer any of them.  (In fact I never had time anyway, but Emily used to be more cooperative.)

So we're not surveyed.

So how common is our reaction? I know we're very peculiar in most things, but I would imagine that the fund raising tricks and phishing exploits must turn quite a few people off of surveys.

PS. We always tell them to take us off the call list, but I doubt that does anything. We're in the 'do not call' registry anyway.

Tuesday, August 24, 2010

Schmidt really did sell Google's soul to Verizon

Apple could do a deal with AT&T and emerge unchanged because, of course, Steve Jobs is Satan. He is beyond mere corruption.

When Google did a deal with Verizon though, they lost everything (emphases mine) ...
Marvin Ammori: Google-Verizon Pact: Makes BP Look Good

A lot of people have been discussing the Verizon-Google pact, including venture capitalists (on NYT's Room for Debate) and Silicon Valley companies. Most people agree: Google does evil, calls it net neutrality.

Last week I wrote up a guide of the FCC negotiations on net neutrality, setting out all the loopholes, and noting that the carriers needed only one loophole to kill an open Internet. Verizon and Google announced their pact two days ago. Rather than including one loophole, they went down the checklist and included just about every loophole they could.

Maybe the most ridiculous one--which has received almost no attention--is something I didn't mention last week. It's the liability limit. The maximum fine for a violation, after all the loopholes are met, is $2 million dollars...
Fixed rate fines are one of the great scams of American government. A true fine would max out at 30% of Google's yearly revenue.

This Taiwanese video is now quite plausible.

Schmidt is a lesser version of Microsoft's Steve Ballmer. If Google doesn't dump him he'll take them down.

Monday, August 23, 2010

The Bowhead whale can live how long?!!

A graphic in Why Can't We Live Forever? (Scientific American) lists maximal lifespan of various familiar species.

The Jellyfish and Hydra are "immortal". The Galapagos tortoise can live 150 years. Humans 122, Dogs 39, Cats 36, Horse 62(!), Chimp 59(!).

There are a couple of surprises there. I didn't dream a horse could live that long. I imagine that's a bit of a problem for many horse owners. I wonder how many horses ever die a natural death, even if the average lifespan is only half that.

And, of course, the Bowhead whale can live up to 211 years.

WTF?! Where did that come from?

I found this 1999 reference, it's apparently the source for the 211 year estimate ...
... Based on these data, growth appears faster for females than males, and age at sexual maturity (age at length 12-13 m for males and 13-13.5 m for females) occurs at around 25 years of age. Growth slows markedly for both sexes at roughly 40-50 years of age. Four individuals (all males) exceed 100 years of age. Standard error increased with estimated age, but the age estimates had lower coefficients of variation for older animals. Recoveries of traditional whale-hunting tools from five recently harvested whales also suggest life-spans in excess of 100 years of age in some cases.
It turns out Bowhead Whales and Naked Mole Rats both have weirdly long lifespans  ...
... Mitoptosis can be also activated in adult postmitotic somatic cells by evolutionary conserved phenotypic adaptations to intermittent oxygen restriction (IOR) and synergistically acting intermittent caloric restriction (ICR). IOR and ICR are common in mammals and seem to underlie extraordinary longevity and augmented cancer resistance in bowhead whales (Balaena mysticetus) and naked mole rats (Heterocephalus glaber)...
So to live long and prosper spend mealtimes holding your breath rather than eating?

Not surprisingly a 2007 article advocating genome sequencing of mammals with atypical longevity ...
... we propose the sequencing of three organisms of unique interest for aging research: the naked mole-rat (Heterocephalus glaber) whose record longevity of 28.3 years makes it the longest-lived rodent (2), the white-faced capuchin monkey (Cebus capucinus) which can live 50 years (3), and the bowhead whale (Balaena mysticetus), the longest-lived mammal with estimates suggesting that it may live 200 years (4).
Hmm. Maybe we should do this sequencing before we drive all whales to extinction?

Even if the 211 year estimate is an exaggeration these whales live much longer than I'd imagined. I'd have guessed 40 years, but they're still young adults at that point.

Until this minute I was skeptical that there was really much room to slow the human aging process. Now it feels inevitable. Not likely in my lifespan, but maybe the future of my children or their children.