The Bayesian Heresy: Understanding Innovation

Scott Page, as quoted by Bayesian Heresy, asserts that a diverse team of good thinkers will be more creative than a less diverse team of very good thinkers:

The Bayesian Heresy: Understanding Innovation

...First, for any problem there exists a perspective that makes it easy to grasp a solution, though that may mean waiting for a person as unique as Edison to come along. Second, across all problems no perspective or no heuristic is any better than any other. In plain English, any approach may be just as good as any other until it is tested.

Third, teams of problem solvers—viewed as bundles of perspectives and heuristics brought together to solve a particular problem—do better when the diversity of perspectives and heuristics is greater than the overall ability or talent of the team’s members. In other words, diverse teams outperform teams composed of the very best individuals. Diversity trumps ability.

This last result requires further explanation. A team, a group, or even an entire society innovates through iterative application of perspectives and heuristics. Individuals who perform best obviously possess good perspectives and heuristics (think Edison), yet 30 Edisons each may have 20 useful heuristics while collectively possessing a mere 25. In contrast, the diverse team’s individual members may on average only know 15 heuristics apiece but collectively know 40.

I don't buy it. There are confounding variables such as communications latency, ability to cooperate, response to incentives, and solution collision (multiple semi-compatible good quality solutions that, when combined, produce a weak result). Diversity is valuable, but I've never seen it trump ability by itself.

The Dreams of Boltzmann Brains

Emphases mine.

OO’s and BB’s | Cosmic Variance

.... The idea Don put forward is this: there’s us, the ordinary observers (OO’s) in the world, who have achieved a certain stature after billions of years of evolution in the universe, and are now capable of making quite refined (or so we think) observations of the universe. ...

Then, though, there are the BB’s in the universe: Boltzmann Brains. Random fluctuations of the fabric of spacetime itself which, most of the time, are rather insignificant puffs which evaporate immediately. But sometimes they stick around. More rarely, they are complex. Sometimes (very very rarely) they are really quite as complex as us human types. (Actually, “very very rarely” does not quite convey just how rare we are talking now.) And sometimes these vacuum quantum fluctuations attain the status of actual observers in the world. But, the rarest of them all, the BB’s, are able to (however briefly) make actual observations in the universe which are, in fact, “not erroneous” as Don Page put it.

The man was a compelling speaker, and soon I realized there was an actual intellectual debate underway in the high end of the cosmology/high energy community as to what the role of these BB’s might be in the universe, in the very far (or maybe not so far) future...

The thing is, when you start talking about very very…very rare things like Boltzmann Brains, you are talking about REALLY long times. Much longer than we’ve had on earth (and I mean 4.5 billion years) by many orders of magnitude. Numbers like 10 to the 60th years were being batted around like it was next week in this talk. By those times, all the stars and all the galaxies have gone out, and gone cold, and space has continued to expand exponentially and things are long past looking pretty bleak for the OO’s still around, who (we presume) need heat and light and at least a little energy of some sort to survive, even if we are talking about very slow machine intelligence (even slower than humans for example).

So eventually, the mere fact that there is, at these long times, just oodles of space in the universe means that the BB’s become more and more common (even if they are rare) and eventually dominate the, uh, intellectual landscape of the universe...

...I am mangling this horribly, and of course before writing this I took just a glimpse at the already voluminous amount of literature on this topic, and realized that I have a lot of reading to do, both blog and academic. So it’s best I stop and let you all go look up Boltzmann Brains, as I will, and do some more reading...

There's a lot of time in a universe that's open (expanding). Time enough for spacetime to dream the lives of mortals ...

Conservapedia: and now for something completely different

Different, as in very funny:

Conservapedia | Cosmic Variance

.... Everyone is having their fun with Conservapedia, a rightward-tilting alternative to Wikipedia that aims to ensure that future generations of conservatives grow up really dumb. A mildly-close look reveals that the major biases of Wikipedia that made this new project worth launching are (1) their insistence in using “CE” (Common Era) rather than “AD” (Anno Domini) in giving dates, and (2) the occasional Anglicized spelling....

CV provides some hilarious Conservapedia entries, please do read the post. At the moment the site is being overwhelmed by parody posts, but I'm sure it will settle down to a duller, but no less foolish, state. The good news is that any migration from Wikipedia to Conservapedia can only raise the IQ of Wikipedia.

Jimmy Carter: can he be made a saint?

Jimmy Carter, age 82, leads an underfunded and increasingly successful assault on river blindness. True, he's not catholic, but couldn't he be made an honorary saint anyway?

I was going to title this "Give the man the damned peace prize already" -- but then I remembered that he finally got the prize in 2002.

Update 2/25/2007: Rumor has it that Carter is pressing Gore very hard to run. I don't America is good enough to merit someone of Al's quality, but we do need him.

How to murder 100,000 people - and get away with it

Counterfeit drugs, mostly manufactured in China, are widely distributed in Southeast Asia and Africa. The thugs have a lot of blood on their hands ...

In the World of Life-Saving Drugs, a Growing Epidemic of Deadly Fakes - New York Times

Estimates of the deaths caused by fakes run from tens of thousands a year to 200,000 or more. The World Health Organization has estimated that a fifth of the one million annual deaths from malaria would be prevented if all medicines for it were genuine and taken properly...

I'm generally not in favor of the death penalty, but it is tempting to make an exception for wealthy entrepreneurs who kill thousands of people.

Which brings me to tobacco. Tobacco executives sell an addictive product that, when used as directed, kills thousands. From a purely ethical perspective, disregarding minor details of law, are they substantially less evil than these counterfeiters?

Krugman has broken me. Oh, and Edwards health plan

A colleague has been dropping off printouts of Paul Krugman's TimesSelect column to my desktop. I'm thereby reminded that there's nobody else in mainstream journalism willing and able to write fact based in-your-face disclosures of what Sauron (Bush/Cheney is the current incarnation) is up to. Krugman is Molly Ivins with less humor but a stronger platform and far more data. Now that Molly has died there's no-one else playing in this league. Alas, the NYT put Krugman behind their $50/year paywall. Conspiracy theorists figure this was a way to silence him, but I think the NYT made a bet-the-ranch decision that people like me would crack -- sooner or later.

Congratulations NYT, you've won. I've cracked. Here's $50. And so, here's Krugman on the Edwards health plan...

Edwards Gets It Right - New York Times

...People who don't get insurance from their employers wouldn't have to deal individually with insurance companies: they'd purchase insurance through ''Health Markets'': government-run bodies negotiating with insurance companies on the public's behalf. People would, in effect, be buying insurance from the government, with only the business of paying medical bills -- not the function of granting insurance in the first place -- outsourced to private insurers.

Why is this such a good idea? As the Edwards press release points out, marketing and underwriting -- the process of screening out high-risk clients -- are responsible for two-thirds of insurance companies' overhead. With insurers selling to government-run Health Markets, not directly to individuals, most of these expenses should go away, making insurance considerably cheaper.

Better still, ''Health Markets,'' the press release says, ''will offer a choice between private insurers and a public insurance plan modeled after Medicare.'' This would offer a crucial degree of competition. The public insurance plan would almost certainly be cheaper than anything the private sector offers right now -- after all, Medicare has very low overhead. Private insurers would either have to match the public plan's low premiums, or lose the competition.

And Mr. Edwards is O.K. with that. ''Over time,'' the press release says, ''the system may evolve toward a single-payer approach if individuals and businesses prefer the public plan.'"

So this is a smart, serious proposal. It addresses both the problem of the uninsured and the waste and inefficiency of our fragmented insurance system. And every candidate should be pressed to come up with something comparable...

This is a great example of why I need Krugman. I skimmed the Edwards plan and came away feeling a bit disappointed. On quick glance it seemed to miss the key issues. Krugman is smarter and looks deeper, and reveals that the Edwards plan is serious and meaningful. Put me down in the Edwards camp -- unless, as Robert Reich half-predicts -- Gore announces his candidacy at the Academy Awards. Then I'd be torn ...

(Now I get to see how much Krugman I can quote before the NYT sends me a nasty-gram. I think I can stay within Fair Use pretty readily.)

How not to thrill your customers

An author of a couple of interesting OS X apps, has a blog. I wrote a (brilliant, of course) comment on the blog -- but I fumbled the captcha. When the screen redrew, my comment was gone.

I figured I'd let the fellow know of the problem, but here's what his blog has in the contact section:

...If you’d like to contact me, please send me email. I’ll leave it as an exercise to the reader to figure out my email address...

Yikes! This guy hasn't figured out that his blog is a part of his sales strategy ... or, in this case, his anti-sales strategy! (Name intentionally withheld.)

Deinventing government: renewing a US child's passport

A golden age can only be recognized in a rear view mirror. Once upon a time Al Gore was reinventing government, and governmental websites were often useful. Alas, in the reign of Sauron all those websites have been outsourced to the highest donor. Consider, for example, the process of renewing a child's passport.

Now, I admit this is an extreme example. A few years ago our very dim congress decided that a child's passport must prove both citizenship and also, somehow, prove that the child is related to their parents. This made the process very complex - for every renewal up to age 14 (after which the child passes into yet another intermediate process). So the process is seriously bunged up to begin with. Even so, the obscure and confusing directions manage to make a bungled process even worse.

For the benefit of anyone who ever has to do this, here's the directions for the worst case scenario (adopted children, mother kept her birth name and is thus surely a terrorist):

1. From Special Requirements for Children Under Age 14 get the forms and print them.
   
   
2. Get pictures at Kinkos. They're open all hours, and they have the right equipment. (Some post offices will do photos! The MSP Airport post office will do the picture.)
   
   
3. Find a post office that does passports. I don't know any way to find the hours they're open for this other than visiting them (!). In MSP (Minneapolis) the airport PO is open for passport processing from 9am-3pm seven days a week and it does photos as well. Other offices may be open on certain hours and days. There's usually no way to phone and discover the hours -- you have to actually make a specific visit to learn the hours.
   
   
4. You will need (for our "worst case" scenario)
   

• Child's current passport (god help you if it's expired)
• Child (in good mood)
• Those pictures from Kinkos, or done at the post office in some cases
  
• BOTH parents physically present, both with identification (I recommend carrying both drivers license AND passport for identification).
• child's certified US birth certificate (every time, you'd think they'd only need it once)
  
• just in case: adoption certificate andmarriage certificate if mother's name doesn't match the children's name. The last time we did this we weren't asked for these.

The passport office will keep the original birth certificate and last passport. We paid the extra $60 for expedited service on the theory that it might reduce the rate of document loss and processing errors. I hate surrendering original documents.

You can check the status of the passport renewal here.

Read the official site for the official list, but the above works for us. I've written my house representative -- at the very least the Post Office hours and service information should be on a web site.

Aetna: winner of this month's stupid security policy

Aetna's username policy requires a number. In the username. On the other hand, they authenticate using the same "security questions" everyone else uses. So if your account is cracked somewhere, the crooks can use your past "security" answers to get your Aetna account too -- even if they don't know your "secret" "username".

Confusing the username with the password. That's rich. Aetna wins this month's stupid security policy prize.

Router admin password: don't use the default!

If you've never set your router's admin (not the wireless pw) pw, you should do it now: Gordon's Tech: The router/javascript bug - this feels big.

Yahoo! has abandoned desktop search

That was quiet! Yahoo! has abandoned the desktop search market. Yahoo! Desktop Search now downloads from X1 -- the company that Yahoo had licensed their product from. The "additional file type" search is no longer free from Yahoo, it now requires an X1 license. (I misread the web page, it's still there!)

This was a big market a few years back, with products from Microsoft, Google, Yahoo, and others. I tried them all, and only Yahoo Desktop Search (X1) was worth the bother. Alas, it's gone now. Google Desktop Search is still around, but it doesn't treat folders with sufficient respect and it doesn't give me enough configuration options.

Microsoft will focus on migrating users to Vista search, and of course OS X has spotlight, so there's only the soon-t0-dwindle XP marketplace for Yahoo to work with. I presume they entered this market in the first place to try to get YDS users to use Yahoo search, but I doubt that was worth the bother. It's especially not worthwhile now that Google has won the search wars.

I may try the X1 product, though when I last tried it I was surprised to find that YDS was less buggy than X1!

Update: I'm using it now. It seems as good as always, but I think YDS might have allowed me to do network shares -- X1 doesn't without an upgrade. In any case I don't need this feature -- my network is mostly OS X.

Update 2: Actually, it's better than YDS. I don' t think YDS included Eudora indexing, but this version of X1 does. Nice!

Four decades of television: last of the rabbit ears

When it comes to television, we're a bit retro. The kids watch Netflix videos (movies and tv) on an ancient CRT, but we don't do broadcast or cable. So when my 10 yo insisted on watching the super bowl while the youngers watched the CRT I had to scramble a bit.

It worked, and I got a kick out of all the decades of technology that were trivially easy to lash together. In an era of Yet Another Proprietary Apple Connector the RCA AV plug is a shining beacon of lost hope.

In front we have a 2006 Dell desktop monitor (HDTV!). The box is a junky DVD/video player with a broken DVD player and a working TV tuner - vintage 2000. Below the table is a vintage 1990 stereo. 1950s era rabbit ears sit atop the box. Fifty years of technology all working together to create a very low end home AV system.

There won't be many more opportunities to put this mash-up together. Broadcast analog television is supposed to end in a year or so, though I personally doubt that will happen. A president may survive the greatest strategic debacle since Japan attacked Pearl Harbor, but ending broadcast TV is a recipe for annihilation.

Cheap stuff that doesn't work: are we figuring this out?

I don't think Apple has particularly good quality assurance. They do, however, have an incredibly strong brand and rabid and demanding user base --- including macintouch. I can wait until the base sorts the flops from the goodies, and then buy with reasonable expectations. Sure, I can still be burned by idiotic proprietary connectors, but at least I don't buy the world's most painful phone or a dysfunctional alarm clock.

Most of the world's gizmos, however, don't come from Apple, Canon, Bose, Nikon, or any other company with a reputation to manage (are there any others?). They come from companies like SONY, Dell, and a large variety of fake brands (Zenith, etc) used by transient contract manufacturers. Some of the stuff works for a while, a heck of a lot doesn't. It mostly ends up in massive piles of toxic junk, and irretrievable hours of lost lifetime. It's a con job, and we've been unbelievably slow to catch on.

Today, though, an unlikely voice of rebellion arose within the very home of the enemy -- Gizmodo (emphases mine):

Horseshoes and Hand Grenades: Joel Johnson Returns...to Spank Us All for Supporting Crap - Gizmodo

....Stop buying this crap. Just stop it. You don't need it. Wait a year until the reviews come out and the other suckers too addicted to having the very latest and greatest buy it, put up a review, and have moved on to something else. Stop buying broken products and then shrugging your shoulders when it doesn't do what it is supposed to. Stop buying products that serve any other master than you. Use older stuff that works. Make it yourself. Only buy new stuff from companies that have proven themselves good servants of their customers in the past. Complaining online about this stuff helps, but really, just stop buying it...

The first step in our rehab is to restore the importance of brands with reputations. There are a few left -- buy from them. Our time is not a commodity.

More evidence of hacked wetware: the DARPP-32 gene and schizophrenia

I think humans have hacked and buggy wetware. I predict that studies underway on the frequency of autism in Ashkenazi populations will show that some of the genes underlying autism also have adaptive advantages.

In the meantime, we may be seeing this in schizophrenia: FuturePundit: Intelligence Boosting Gene Ups Schizophrenia Risk. The DARPP-32 gene provides a cognitive boost, but may also increase the risk of developing some forms of the mixed set of disorders we call "schizophrenia". This is what you see with "new" mutations that haven't been fully debugged yet.

The sooner we understand how buggy all our minds are, the sooner we'll learn wisdom, tolerance, and forgiveness.

Quantum computing: revenge of the analog

I'd assumed we were still 10 years away from a useful quantum computer. A Canadian company is now advertising one: Scientific American.com: First "Commercial" Quantum Computer Solves Sudoku Puzzles.

The experts believe this particular approach won't scale, but I'm stunned that they're already up to 16 bits. It also doesn't work do encryption tasks, so international finance can keep working. It apparently works rather like the analog computers my grad school adviser once played with. Don will be amused to learn that his analog skills are again on the cutting edge.

So, how long until we do applied counterfactual computing? Does anyone still think we can predict twenty years ahead?