Saturday, March 21, 2009

Skynet cometh

A dimly light cubicle, late in the night. Three large monitors display a bewildering array of controls and monitors, and Google Reader.

"Hitting the blogs Mike?" said Teresa as she sat on the table edge.

"Yeah, my brain is fried. We've tried everything, but all we're getting is basic AI. Sergey's gonna send me to tech support if we don't get something."

Teresa leaned forward. "Hey, what's that article?"
The human brain is on the edge of chaos - Follow Me Here…

“Cambridge-based researchers provide new evidence that the human brain lives “on the edge of chaos”, at a critical transition point between randomness and order. The study, published March 20 in the open-access journal PLoS Computational Biology, provides experimental data on an idea previously fraught with theoretical speculation...

... According to this study, conducted by a team from the University of Cambridge, the Medical Research Council Cognition & Brain Sciences Unit, and the GlaxoSmithKline Clinical Unit Cambridge, the dynamics of human brain networks have something important in common with some superficially very different systems in nature. ... critical systems are able to respond very rapidly and extensively to minor changes in their inputs.”
"Hmm", said Teresa. "You know, if we crossed the signal inputs with a chaotic amplifier we could get something like that ..."

"Damn, I'll give it a try. Not too hard ... there. It's done. Now we wait."

"I'm hungry. Pizza?"

"Sure. Oh wait. What's that?"

"Looks like a firewall down Mike. That's weird. Hey, there's another. Hell, they're all dow..."

The senescent web - lessons from eHow's cedar shake advice

In the middle age of the web there are an astounding number of garbage web sites. I fear the ratio of garbage to genuine sites is now approaching ten to one on Google search pages for some topics.

Consider home repair. As near as I can tell several businesses have created junk-bases full of stolen or ridiculously lousy advice. They then publish this under hundreds of domains, and sell ads across all of them. Many of these sites show up in Google searches, so today they're winning the algorithm wars.

I recently read a science fiction writer's description of a future "web" that was 99.99999% garbage, some of it insanely subtle garbage. None of this was apparent to users, because near-sentient AIs filtered 99.99999% of the garbage out.

A completely unrealistic scenario of course, since any such AI would be fully sentient and quite uninterested in such a menial task.

The web, inevitably, is going the way of usenet. It's being overwhelmed by fraudulent junk. It's barely twenty years old, and it's already senile.

There's hope, however. We carry a huge amount of noise and parasitic junk in our DNA but complex life forms exist. From chaos can arise a metastable system.

Maybe, on the way to Skynet, the web will do the same. In the meantime, on the web, as in the real world, reputation and brand identity matter more than ever.

When I was looking for advice on repairing dog-broken cedar siding shakes, eHow, a relatively old and established site, had the best advice. It has ads too, but there's a reasonable balance between ads and content.

There's a lesson here for Google, Yahoo, and especially for Microsoft. Google's algorithmic approach to search is quite vulnerable to fraud. In some domains it's failing. On the other hand a simple-minded Yahoo-1996 style approach is vulnerable to bias and other kinds of deception. If Microsoft can find a way to solve this particular problem, by joining algorithmic and human judgment, they may yet challenge Google.

On the other hand, if Google solves it ...

Friday, March 20, 2009

Google GrandCentral (Voice) has been saving me $80 a month

I started using GrandCentral last August, many months after I’d signed up for the service.

At that time, post acquisition, Google wasn’t marketing GrandCentral, so it took me an inordinately long time to catch on to the real value. I could call my aged parents in Montreal for free.

Now I’ve switched to Google Voice, and my calling cost has risen astronomically – to 1 cent a minute [1]. With the switch, and some minor glitches in the world economy that incent savings, I decided to look back at our cell phone record and see how much GrandCentral was saving us.

Eighty dollars a month.

Wow.

Ok, so I’m an extreme case. I call my mother on my commute home (it’s not very distracting, honest) so the time adds up.

Still, that’s a lot of money. It not only pays for my iPhone data plan, it’s now covered the cost of my iPhone. Not that Google Voice requires a smart phone (though I sure miss GrandDialer), I can make the calls from any phone for the same price.

Google Voice will be available to everyone in the US shortly [2]. Around the same time iPhone 3.0 will give me free instant messaging (background push notification).

I’m surprised AT&T’s share price hasn’t started falling.

[1] Precisely 1 cent a minute. None of the tricks those asinine international plans of old used to play. A 22 minute call costs 22 cents, a 1 minute call costs 1 cent.

[2] Looks like it may launch in Canada around the same time, other nations at varying times. It will make a big difference for some family members of ours.

Update 8/09: Google Voice calls to Canada are again free.

Fastest IE 8 bug find ever

I’m cursed.

I use Chrome 2.0 beta for 1 minute, and I find a significant bug.

I buy the iPhone and spend months raging a personal war against bugs and missing functionality while the world celebrates.

My Google Cloud torments me. At one time last week I was at war with Google Gmail Video Chat (unstable), Google Reader (team was at a conference, so they missed a significant function outage for 6 days) and Google Outlook Sync and Google iPhone sync (significant bug, I found a fix).

Let’s not talk about Windows, shall we. Or Apple iChat. Or the botched OS X model for file permissions. Or iCal. Or …

Cursed.

Even by my standards though, Internet Explorer 8 install was impressive. I found a significant bug before it installed!

The benighted installer copied the install files to my “D:” drive – which does exist.

Why the D: Drive?!?!

Well, I saw something like this before with Microsoft Office installers in 2006

… Use Regedit to find "LocalCacheDrive" settings for Office. Notice that the drive letter is "D" when it should be "C"…

…In retrospect I think this all began when I installed Office with the 'remove install file option'. I use that option because I kept my install files on a hard drive. Alas, a separate hard drive in those days. Drive letter "D". Nowadays drive letter D: does not exist. The bug bites when Office looks for its install files using a drive letter that no longer exists. Office doesn't produce a dialog box or a reasonable error message, it just dies. The automated install process persists daily in the futile update. (Update: I've been told this was an ancient bug with Windows update too.)…

Yes, IE 8’s installer has the same I ran into in 2006, a bug that was very old even back then.

Update: Well, not quite the same bug. When I turned off drive D: there was a delay of a minute or so, then the installer continued using drive C: instead.

Of course on restart I had no network connectivity at all on my XP machine ...

Update: A restart cured the network hang -- looks like it was the result of having some Microsoft security updates waiting to be installed on restart. IE 8 is definitely snappier than IE 7 on my old XP box, but I have to run it in compatibility mode to get Blogger to render correctly.

How should we treat scientific fraud?

We know there's a lot of fraud on Wall Street. It doesn't end there though ...

Anesthesiologist Faked Data in 21 Studies: Scientific American

Over the past 12 years, anesthesiologist Scott Reuben revolutionized the way physicians provide pain relief to patients undergoing orthopedic surgery for everything from torn ligaments to worn-out hips. Now, the profession is in shambles after an investigation revealed that at least 21 of Reuben's papers were pure fiction...

... Reuben, 50, has been stripped of his research and educational duties and has been on medical leave since May...

... His lawyer, Ingrid Martin of Dwyer & Collora, LLP, in Boston, told ScientificAmerican.com that Reuben has cooperated with the investigation and that he "deeply regrets that all of this happened." She added that "with the [investigating] committee's guidance, he is taking steps to ensure this never happens again."

... Reuben's career would begin to unravel as Ekman began to suspect foul play. In addition to collaborating with Reuben on the now-retracted Celebrex study, Ekman agreed to review a Reuben manuscript on surgery on the anterior cruciate ligament (ACL) in the knee. But when he asked the anesthesiologist for the name of the orthopedic surgeon on the study, Reuben ceased communication with him...

... By then, Editor in Chief Shafer had already put several Reuben manuscripts on hold after learning that Baystate had initiated a probe into the validity of his research. The investigation later identified 21 articles based on patient data that had been partially or completely doctored...

This man caused great harm. He effectively stole large amounts of money. His fake results misled researchers, clinicians and patients and likely led to at least some suboptimal care.

He's done vastly more harm than the average crack addict, but in some states someone caught using crack cocaine can spend years in prison.

Why don't people who commit scientific fraud go to prison?

Another day, another 200 billion to pay

More taxes to pay one day for the trillion printed today ...
Fiscal aspects of quantitative easing (wonkish) - Paul Krugman Blog

... My back of the envelope calculation looks like this: if the Fed buys $1 trillion of 10-year bonds at 2.5%, and has to sell those bonds in an environment where the market demands a yield to maturity of more than 5%, it will take around a $200 billion loss.
It's comforting that we'll pay I guess. I assume the interest rate more or less works out to something near the Fed rate, and that Krugman has corrected for inflation with the loss amount.

Let's assume there are 100 million Americans who'll be paying back the $200 thousand million. That's $2,000 apiece, more for high earners so say $6,000 if you're doing ok.

That's a lot of money. On the other hand, if I'm unemployed for a few months I'm out far more money than that. Adjusted for risks and so on it seems like a reasonable bet even for those doing relatively well.

For folks not doing so well it's a bargain. Of course if the market were to recover it would be a real bargain for us.

Thursday, March 19, 2009

Religion and recycling bin aversion

We all know we're heading for the recycling bin. (Yes, you too Mr. Kurzweil.)

Beyond that, many Americans say they expect better things. Yet they don't act on those stated beliefs ...

Religion, medicine and evading death | But not yet, Lord | The Economist

HOW do a person’s religious beliefs influence his attitude to terminal illness? The answer is surprising. You might expect the religious to accept death as God’s will and, while not hurrying towards it, not to seek to prolong their lives using heroic and often traumatic medical procedures. Atheists, by contrast, have nothing to look forward to after death, so they might be expected to cling to life.

In fact, it is the other way round...

So does religion promote fear of death, or does fear of death promote religion?

My money is on the latter.

Jurassic Lark? Nooooooooo

First Pluto. Now T Rex?

Did all dinosaurs have “feathers”? | Our feathered friends | The Economist

... Taxonomically, the very definition of a bird was until recently an animal that has feathers. Now, taxonomists argue that since birds are descended from dinosaurs they should be classified merely as a subgroup of the Dinosauria. But if feathers truly are the diagnostic criterion, then perhaps things should be the other way round, and Stegosaurus, Triceratops, Diplodocus, Tyrannosaurus and their kin should no longer be thought of as terrible lizards, but as overweight, flightless birds.

It must stop now.

Urban renewal - Habitat tears down houses

I think this is a good sign ...

Saginaw Journal - In Hard-Hit Areas, Habitat for Humanity Adds Demolition to Its Mission - NYTimes.com

... International leaders of Habitat for Humanity, an organization more than three decades old, say their focus is changing to meet the demands of a changing economy. In cities where so many homes sit empty, the group is leaning away from building new houses and instead fixing up old ones, said Ken Klein, the vice chairman of the group’s board.

In recent years, about 100 of the organization’s affiliates around the country have done the same, removing recyclable items, like cabinets, floorboards, plaster and light fixtures, from condemned houses and, in a few cities, even razing some structures...

When we drove across the northern-eastern border of the United States last year I was forcibly struck the immense stretch of desolation.

The region has been in decline for decades, but things felt much worse than even two years before. Since then we've been reading reports of vast stretches of Cleveland with many abandoned homes.

People still live where the lumber, water, and railways of the 19th century put them. In some cases new economies make those places viable. In most cases they, like the American automotive industry, need to downside by at least 50%.

It's possible to image a satisfactory end point in a world where telepresence and telecommuting are becoming commonplace. Cities with large parks and green spaces, better aligned along transportation corridors. Larger homes with expansive lawns. Bicycle paths and playgrounds.

To get there however a lot of homes have to go, and existing homes must be refurbished and expanded.

The classification and heritage of American torture

Everything has a history...

The history of CIA torture - By Darius Rejali - Slate Magazine

In the 20th century, there were two main traditions of clean torture—the kind that doesn't leave marks, as modern torturers prefer. The first is French modern, a combination of water- and electro-torture. The second is Anglo-Saxon modern, a classic list of sleep deprivation, positional and restraint tortures, extremes of temperature, noise, and beatings.

All the techniques in the accounts of torture by the International Committee of the Red Cross, as reported Monday, collected from 14 detainees held in CIA custody, fit a long historical pattern of Anglo-Saxon modern. The ICRC report apparently includes details of CIA practices unknown until now, details that point to practices with names, histories, and political influences. In torture, hell is always in the details...

...For 30 years, I've studied a long and remorseless two centuries of torture around the world, and I can find only one instance of an account resembling the collars and plywood technique described in the ICRC report...

America needs a truth commission.

The other side of a bad kid dead

A troubled young man with a record commits suicide in jail.

Just another local story.

Except there were a few people who loved him.

There usually are.

Against creationism - the toenail

No rational creator would give humans toenails.

Ouch.

Wednesday, March 18, 2009

Get ready to refinance your mortgage?

The Fed just “printed” a trillion dollars.

One analyst expects mortgage rates to fall as a result …

Economists’ Reactions to the Fed’s Move - Economix Blog - NYTimes.com

David Greenlaw, Morgan Stanley: The Fed’s announcement signals a clear intent to continue to drive mortgage rates lower and we expect them to meet this objective. This could represent a powerful source of stimulus for the household sector of the economy. In 2008, the average mortgage rate on the outstanding stock of loans was about 6.50%. So, if the Fed brings 30-yr fixed rate mortgages down to 4.50% and all homeowners are able refi, the aggregate permanent cash flow savings would be on the order of $200 billion per year.

We’ll be watching.

Of course now we’ll all be watching for bubble #3. We had stocks, then we had real estate. What’s left? A bonds bubble?

Annals of poorly chosen words - best and the brightest

I haven't had a strong opinion on what the feds should do about the AIG bonuses you and I are paying to the team responsible for history's greatest financial disaster. It feels like only one manifestation of a much bigger failure of the compensation market, a zero-bound type failure because the winners now work to win, not to live.

In other words, the winners of the compensation wars can leave the game whenever they want. They play by different rules than the rest of us. We can change that in the future, but it's hard to fix it today.

David Leonhardt's observation does a nice job of skewering one common rationale however ...

Ah, retention pay. It has been one of the great rationales for showering money on chief executives and bankers regardless of how well they are doing their jobs. It’s just that the specific rationale keeps changing.

In the booming 1990s, companies supposedly had to pay retention bonuses because executives had so many other job opportunities. There was a war raging — a war for talent, said McKinsey & Company, the consulting firm.

Then came the aftermath of Enron, when new scrutiny and regulations apparently made some chief executives wonder if they still wanted their jobs. “I’m thinking of actually getting out,” David D’Alessandro, the head of John Hancock Financial Services, reported hearing from one fellow chief executive. The antidote to such doubts? Retention pay, obviously.

Now comes Mr. Liddy, the government-appointed chief of A.I.G., defending multimillion-dollar bonus payments for the people who run the small division that brought down the company. If the government doesn’t let them have their money, they will walk away, Mr. Liddy says, and nobody else will know how to clean up their mess...

So this is a big and deep problem, but skewering AIG's winners is most likely a distraction. We've been taken, but if we spend all our energy beating up on these particular winners we're likely to forget about the bigger failure in the global compensation market.

On the other hand, I do have a strong opinion that Edward Liddy, the winner who returned from retirement to manage AIG [1], showed the insight of a gnat when he wrote (italics mine) ...

... We cannot attract and retain the best and the brightest talent to lead and staff the A.I.G. businesses — which are now being operated principally on behalf of American taxpayers — if employees believe their compensation is subject to continued and arbitrary adjustment by the U.S. Treasury...

Even in the context of our day, that's breathtaking. It's so revealing that Liddy could write that, and that Larry Summers didn't blow a gasket reading it.

Maybe it's just basal ganglia running on automatic. CEO's everywhere call their team the "best and the brightest" even when they've already decided to fire everyone and try with a new batch of losers. In which case using the phrase in this context is merely an idiotic mistake.

It's more likely, however, that Liddy believes he and his team are indeed "the best and the brightest". Winning does that to primates, it's in our genes and our neurochemistry. Unfortunately, Geithner and Summers may believe it as well.

Whatever else comes of this bit of interesting times, let's skewer, rend and bury the phrase "best and the brightest" for all time. CEOs, find another euphemism for "you're all a bunch of losers but I can't get rid of you today".

[1] Update: Turns out if AIG goes bankrupt Liddy loses a massive retirement package that he's probably depending on. So he does have skin in the game.

Update 3/19/09: I think this is right ...

Why AIG paid the bonuses (Nate Silver, 538)
The thing about these "bonuses", however is that they're not really bonuses, which we usually think of as incentive-based compensation. On the contrary, they are something the opposite of bonuses: they took compensation that had been incentive-based and guaranteed it...
The fundamental issue here what I call asymmetrical agency bias. We as human beings tend to attribute our results to skill when we are performing well, but (bad) luck when we are performing poorly...

... I'm interested in compensation and incentivization structures in general. Aggregate compensation throughout the financial services industry, I would guess, is much higher than is economically optimal (there is a lot of evidence that this is true of CEO pay). A lot of people are getting paid for what is thought to be skill but is really just luck (or economic rent).

If, as at most hedge funds, the employees are buying in with their own capital and bearing a lot of the downside risk, that is one thing. At a publicly-traded company, however, those employees are taking profits out of the shareholders' hands. And at a publicly-traded company that happens to be owned by the taxpayers, they're taking money out of the taxpayers' hands.

The compensation paid to AIG's employees, however, is less a moral failure than a market failure. We don't like to admit to market failures because they indict our collective judgment; instead we scapegoat and move on. But there are some ways to address these market failures; the more time we spend focusing on those, and the less on AIG, the more money we the taxpayers will save ourselves in the end.

Tuesday, March 17, 2009

Avoiding the baby in car tragedy - the tethered parent

When I read of a parent who forgets their infant in car seat and leaves them to die in a hot car, I don't think "bad parent". I think "thank you random chance, there but for the odds go I".

I remember a young child asking me from his car seat why I'd missed the turn off for work. (This was a "special needs" child who's always had a knack for directions.) I'd forgotten he was there. Bad feeling.

About 20 American parents (mostly fathers I bet) make the ultimate mistake every year. Typically it's a change to an automatic routine, a chance request to drop a child in day care on the way to work. The infant falls asleep, the basal ganglia takes over, then tragedy.

My children are old enough that they can get out of the car. If they were in car seats though, I'd follow this advice when in the high risk situation of deviation from a common habit ...
Schneier on Security: Leaving Infants in the Car (comments)

... Secure a string to the car seat and yourself (belt, wrist, etc.). Leave plenty of slack so it doesn't interfere with driving, but if you try to exit the car and forget, the string will tug. Having to unclip the string will provide the reminder...