Stephen Baker

I'm thinking of buying an American car, preferably one produced by Michigan's battered industry. The new Ford Focus gets great reviews, so yesterday I visited one of my local New Jersey dealers. Now when my wife shopped for her European-made Mini Cooper, we went to the wealthier suburbs where the salespeople seemed to know our lives. ("The latte won't spill if you put it here....") But when I want to buy an American car, it turns out, the salesman I deal with is foreign born. We don't communicate all that well. This is by no means a deal-breaker, but I do find it interesting that I have to bridge a cultural gap to buy an American car.

One problem for me is that he's willing to give me only $800 for my trade-in, my still roadworthy 1997 Nissan Maxima (with 158,000 miles on it). In the classifieds, it looks like people are asking about $4,000 for comparable cars. (I'd sell mine for $2,500 in a minute, if you're in the market.)

If I go ahead with this purchase, it'll be my first new car. It would assume its place in a lineage that's by no means noble. My car history:

1978: Bought a '72 Saab 99E, for $1,700. This would be my car for my reporting job at The Black River Tribune, in Ludlow, VT. I'm sure I abused it. It broke down all the time. I headed off to New York in 1980, resolved never again to buy a car, but I didn't picture myself in Caracas.

1984: Purchased a 1978 VW bug in Venezuela for about $1,000. This car forded rivers, climbed into rain forests, jostled bravely in Caracas traffic. Sold it for $700 when I left a year later.

1985: Bought a 1980 Plymouth Horizon, a cheap and tinny Japanese import for $1,700 in Portland, OR., and drove it to my new job in El Paso, TX. It overheated in Nevada. To keep it from melting, we had to drive the rest of the way by night. It went downhill from there. Unloaded it for $600 in late '86.

1987: Paid $950 for a 1965 Plymouth Valient to a New Mexico farmer, and drove the car to my new BusinessWeek job in Mexico City. I quickly replaced the 8-track tape deck with a cassette player, and eventually installed seat belts in the back seat. This was the closest I've come to loving a car. Once in Mexico City, a cop stopped me just to ask if he could buy it. I drove it up to Pittsburgh in 1992, but it didn't pass inspection. Sold it to an antiques buff for $500.

1965 Plymouth Valient (The "Val') photo from CarGurus.com

1992: Bought a 1981 Buick Skylark from a mechanic in Pittsburgh for $1,500. After the '65 Val, it felt like a Cadillac. Power steering, air-conditioning, cruise control, and surprisingly good gas mileage. Under its thrall, I wrote an article for BusinessWeek about the joys of driving a junker. The lead:

On summer afternoons, we would park the battered '81 Buick Skylark between home plate and the poison-ivy patch. It was the ideal backstop, and it came with air, cruise control, even a decent stereo. For a while, the kids complained about the unglued ceiling, which draped low, making the car feel like the inside of a tent. That was easily fixed with a staple gun. And when it came to vacations, this Buick was perfect. If it broke down, we would simply junk it and rent another. No headaches with Cape Cod mechanics. We owned a disposable car.

Just reading that old article has me thinking that I should hang onto the Maxima... In any case, when that Skylark finally died...

1997: I bought a joyless and underpowered 1985 Toyota Camry for $2,000, and sold it for that same amount a year and a half later when we left for Paris. (Toyotas hold their resale value, I learned.)

2000: Splurged on a 1997 Peugeot 406, spending $10,000. (The dotcom boom was making me feel rich.) It might have been the best car I ever had. I imagined it felt like a Mercedes. It glided along Paris boulevards and effortlessly cruised at 100 mph on German autobahns. When the dotcom boom crashed, BusinessWeek yanked me back to New York, and I had to unload the Peugeot in a hurry, for $6,500.

2002: Newly arrived, we were sitting at a Starbucks in Montclair, when we saw a 1997 Nissan Maxima across the street, a For Sale sign in the window. We bought it later that day for $8,700. It had about 70,000 miles on it. It has 158,000 now, if you're interested.

1 comment [view]

George Dyson on computing "biology"

February 20, 2012

George Dyson, author of the 1998 classic, Darwin Among the Machines, is coming out with a new book, Turing's Cathedral, the Origins of the Digital Universe. For those who enjoyed the story of Watson, this is bound to be interesting. Dyson discussed some of the issues in a Q&A with Wired's Kevin Kelly.

Some highlights:

On the growth of digital "organisms:"

Besides obvious ones like computer viruses, we have large, slow-moving megafauna like operating systems and now millions of fast-moving apps, almost like microbes. Recently we’ve seen enormous conglomerations of code creeping up on us, these giant, multicellular, metazoan-level code-organisms like Facebook or Amazon. All these species form a digital universe.

The speed of the digital universe's expansion:

Like our own universe at the beginning, it’s more exploding than expanding. We’re all so immersed in it that it’s hard to perceive. Last time I checked, the digital universe was expanding at the rate of 5 trillion bits per second in storage and 2 trillion transistors per second on the processing side.

Alan Turing's "decision problem" and its consequences:

It can be stated as: Is there a formula or mechanical process that can decide whether a string of symbols is logically provable or not? Turing’s answer was no. He restated the answer in computational terms by showing that there’s no systematic way to tell in advance what a given code is going to do. You can’t predict how software will behave by inspecting it. The only way you can tell is to actually run it. And this fundamental unpredictability means you can never have a complete digital dictatorship with one government or company controlling our digital lives—not because of politics but because of mathematics. There will always be codes that do unpredictable things. This is why the digital universe will never be a national park; it will always be an undomesticated, unpredictable wilderness. And that should be reassuring to us.

Big Data and math

February 12, 2012

Datamining

Steve Lohr has a good round-up of Big Data trends in the Times. It has very similar themes to The Numerati and to Ian Ayres' SuperCrunchers.

In fact, reading the story reminded of my BusinessWeek story, Math Will Rock Your World, which led to The Numerati. It argues the same points, but instead of focusing on the subject of the investigation--data--it looks at the tools employed, mathematics and computers (without, you might note, shedding any light on how the mathematicians and computer scientists carry out this work.)

As I've mentioned before, Steve Adler, the editor in chief, started the process by telling me to write a cover story on math. So I started interviewing mathematicians. I was learning all sorts of interesting things about encryption and operations research, but I didn't really see the BusinessWeek cover story until I delved into the world of data. In the end, I wrote a story about Big Data--but kept math in the headline.

A few paragraphs from that story:

The world is moving into a new age of numbers. Partnerships between mathematicians and computer scientists are bulling into whole new domains of business and imposing the efficiencies of math. This has happened before. In past decades, the marriage of higher math and computer modeling transformed science and engineering. Quants turned finance upside down a generation ago. And data miners plucked useful nuggets from vast consumer and business databases. But just look at where the mathematicians are now. They're helping to map out advertising campaigns, they're changing the nature of research in newsrooms and in biology labs, and they're enabling marketers to forge new one-on-one relationships with customers. As this occurs, more of the economy falls into the realm of numbers. Says James R. Schatz, chief of the mathematics research group at the National Security Agency: "There has never been a better time to be a mathematician."

From fledglings like Inform to tech powerhouses such as IBM (IBM ), companies are hitching mathematics to business in ways that would have seemed fanciful even a few years ago. In the past decade, a sizable chunk of humanity has moved its work, play, chat, and shopping online. We feed networks gobs of digital data that once would have languished on scraps of paper -- or vanished as forgotten conversations. These slices of our lives now sit in databases, many of them in the public domain. From a business point of view, they're just begging to be analyzed. But even with the most powerful computers and abundant, cheap storage, companies can't sort out their swelling oceans of data, much less build businesses on them, without enlisting skilled mathematicians and computer scientists.

The rise of mathematics is heating up the job market for luminary quants, especially at the Internet powerhouses where new math grads land with six-figure salaries and rich stock deals. Tom Leighton, an entrepreneur and applied math professor at Massachusetts Institute of Technology, says: "All of my students have standing offers at Yahoo! and Google." Top mathematicians are becoming a new global elite. It's a force of barely 5,000, by some guesstimates, but every bit as powerful as the armies of Harvard University MBAs who shook up corner suites a generation ago.

Algorithms reflect racial stereotypes

February 9, 2012

Yesterday, I joined the 5 million+ others who have watched President Obama singing a snippet of Al Green. With that music in my head, I called up more Al Green on my iPod as I began my workout, and then I hit the little "genius" logo to create a song list along that theme. This brought up songs from The Four Tops, Stevie Wonder, The Staple Singers, Aretha Franklin, and lots of others that fall under the rubric R&B.

When Apple introduced its genius service, I wrote a post for the old Blogspotting blog asking if the service extended racial groupings and stereotypes into the realm of algorithms. My point was that if a musical anthropologist were to put Al Green songs into a list, he or she might find surprising similarities of rhythm, themes, falsetto, with music produced by a wide range of musicians, not all of them of the same race. A first version of the post, which is linked to there, was insensitive in its wording, and it got a lot of people angry. (Questions of racism tend to do that, especially when the potentially offending party is Apple.)

My conclusion isn't that complicated. The Genius service, and others like it, get their data from us. And so if they put Al Green with the Supremes and not, say, with the BeeGees, it's because they're following the public's patterns. (If you think my hypothetical is off-base, I should note that Al Green did a great cover of a BeeGee's song, linked below. Not surprisingly, when I went to it on YouTube, an Obama for America add popped up right next to it.)

The statistical upshot, I would argue, is that the most common grouping listeners find for Al Green is not 1) religious, 2) love songs, 3) 70s pop, or 4) vocal genius, but instead, 5) black. And when you see the vitriol directed at Al Green's latest imitator and admirer, President Obama, you'd have to conclude that many view him first and foremost the same way.

A restaurant's trivia quiz ... evolves

February 7, 2012

I recently heard from George Healy, manager of Maria's Italian Kitchen, in Encino, Calif. For years, he's been running trivia contests in the restaurant. He writes:

Google (primarily but not exclusively, of course) has loomed larger and larger over the questions as time has gone on. The percentage of people guessing as opposed to de facto researching over clams linguine has plummeted over to almost nil (excepting children). And the outrage of people who can't find the answers to questions via Google or Siri has become almost confrontational, almost venomous.

Years ago, when people were 'phoning a friend' to answer a question, we would meet people, have conversations, even find people who WERE the answer to said question. But the lion's share of the conversation now is the validity of the answer if it ISN'T on Google, how could Google have failed, etc.

Maria's Italian Kitchen

While researching Final Jeopardy, I encountered lots of people who struggled to understand what was different about Watson, the Jeopardy computer, and seemed to believe that Google already answered everyone's questions. And yet, as George's note makes clear, people are often frustrated by search engines. Web search is a part of their "brain" that often doesn't work that well.

A few days later, George wrote back with an anecdote:

Years ago my dad called me to offer a trivia question for me to use in the restaurant. It was 'what boxer fought for a championship in their very first professional fight?' I told him that was a great question... then asked for the answer. He said he didn't remember. Arrgh. So i went online and searched whatever preceded Google and - nothing. I looked in books - nothing. I put it up at the restaurant and nobody knew. I called other managers (we had 9 locations) and asked them to put it up, ask sports fanatics - nothing. After a couple weeks i had given up. Then a manager called me to tell me the answer: Pete Rademacher. I was stunned. What an obscure name. I asked the manager how he got the answer and he said: 'Pete's standing right next to me.' He was a regular (once a week) customer who had been coming in for years.

After receiving George's email, I typed queries into Google and Bing to see if they could find Rademacher. It was really hard. "Boxer fought for championship first fight."...."First pro fight for boxing title" etc. I failed to come up with good queries, and Rademacher's name didn't show up.

Now I can't guarantee that Watson would have gotten that one right. The computer, after all, never ceased to surprise me. But I'd put money on it.

1 comment [view]

Driving without GPS

February 5, 2012

A month ago, on a drive to New Hampshire, I tried something so old it was new. Sitting in the navigator seat, I left my GPS-equipped smart phone in my pocket and instead opened an old Rand McNally road atlas. It turned out to be the easiest, most hassle-free drive in a long time. A New York Times opinion piece about GPS and the brain reminded me of that trip. Since then I've been wondering why the old-fashioned map proved to be so calming.

My first thought is that GPS delivers too much detailed information. For a single exit, for example, it might tell you to turn right and continue for 650 feet, then take another right, continuing for 800 feet, and then a left. So you're looking from your phone to the road, asking: "Have we gone 650 feet yet? Was that the right?! Each step of the instructions raises a new question, and insecurity. The map, by contrast, simply shows you that in Hartford you switch from I-84 to I-91 N. That step might involve three or four smaller turns, but they're all well marked and relatively easy to follow. For this, we rely on our eyes and navigations skills we've developed for decades, or even centuries, and there's no glitchy machine interface to deal with. I can't emphasize enough how effortless it was.

Google, the same company that makes the popular mapping app in my phone, is also developing robotic navigation for cars. And if you think about it, the GPS instructions are perfect for the bots. Those machines aren't nearly as skillful as we are at looking around, picking up landmarks and reading signs. But tell them to make three right turns and a left at precise distances, and they dutifully process the commands. So, in that sense, those of us who use GPS are repressing the human skills that computers struggle to match. Instead we mold our minds to a stream of data generated by and for computers. Trouble is, when it comes to robot skills, we're mere apprentices. A Google car "knows" that it has traveled 837 feet. We don't, and it provokes anxiety in many of us.

(continued below huge map illustration)

(photo from BigStock.com)

The easy conclusion to draw would be that the natural me, my senses in harmony with the map on my lap, was navigating the drive to New Hampshire. The GPS-wielders, by contrast, are denatured humans who turn their back on rich analog skills and, as the Times article suggests, stunt navigational development in the brain.

However, it's worth nothing that at one point in human development, maps were new. Back then, I'm sure, similar debates flared. Old-fashioned orienters relied on smells, winds, landmarks, and above all, stars, to find their way. Those wielding maps were substituting new-fangled symbols for old-fashioned data, and they risked sacrificing ancient skills that humans had been developing for eons. By picking up maps, we turned our back on nature. It's a process we've been following for at least 50,000 years.

In his captivating book, The Tiger, John Vaillant argues that the first form of human literacy was our ability to "read" the trail of the animals we hunted. He writes:

The first letter fo the first word of the first recorded story ever was written--"printed"--not but us, but by an animal. These signs and symbols left in mud, sand, leaves, and snow represent proto-alphabets. Often smeared, fragmented, and confused by weather, time and other animals, these cryptograms were life-and-death exercises in abstract thinking.

He goes on to say that we learned to track from the animals ourselves, back when we were closer to those animals in every way than we are now. In a sense, like Rudyard Kipling's Mowgli in The Jungle Book, we were schooled by other animals. They taught us how to read. We shared their food, their environment, and many of their skills, and then we shed those skills as we developed new tools of our own. Some of us return to the woods, or even take classes, to regain a measure of that lost knowledge. But most of us simply move on.

GPS marks another graduation, and delivers another set of once-vital skills to hobbyists. And the only reason that that map felt more comfortable on my lap was that GPS is still in its early days. For now, it's more attuned more to Google's robotic cars than to humans. That will change, and so will we.

3 comments [view]

Watson as a marketer: my upcoming talk

February 2, 2012

Jeopardy book

When machines like Watson start showing up in marketing departments, which jobs will they handle? More specifically, which pieces of work will they handle, and transform? I'll be speaking about this on March 5, at Brite '12, a conference at Columbia University on brands, technology and innovation.

I haven't figured out exactly what I'll say, but the focus will be on language. If marketers with no programming or data-mining know-how can ask the machine in natural language to look for something, that Watson-like machine, using its language skills, will be able to scour not only traditional data, but also thousands of written analyses. Then, using a combination of its language smarts and marketing analytics, it should be able to come back with a list of suggestions, or hypotheses, about marketing campaigns that might work.

I know there's a lot of skepticism that machines can handle such work. It's a different problem than Jeopardy. But for me the question is not if, but when. Think about it this way: Can you imagine in, say, the year 2030 that machines will not be handling this work? I can't. So then you just have to draw the line backwards in time, toward now. Yes, in these early years, a language-savvy marketing machine is sure to misunderstand queries and botch lots of the results. Skeptics will laugh. But with machine learning, the machine will improve, inexorably, convulsing an industry as it does. That's how these machines progress. I witnessed it with IBM's Watson. (For a speculative look at the future of this and other technologies, check out this graphic from Envisioning Technology 2012.)

***

We were down in Ft. Myers/Sanibel Island over a long weekend. In Ft. Myers, we went to Six Mile Slough (pronounced "slew"). It was a beautiful walk. My only complaint (a constant for me in modern life, and especially in Florida) is that the nearby traffic was loud. We walked on a boardwalk through wetlands, saw birds, a small alligator, and lots of beautiful reflections in the shallow mangrove pools. It struck me, looking at the photos, that through the years I've turned things around. I used to look at abstract art and try to see what they were representing. Now I find myself looking at the world and trying to find the abstractions. Does that happen to you?

In Sanibel, we went to Ding Darling Wildlife Refuge. Don't miss it if you're down there, and try to go near dawn and/or sunset.

And then you drive back to the Florida most of the humans live in, the burbs...

A dating app for when things don't click

January 24, 2012

When I was working on The Numerati, my editor was emphatic about one thing: There had to be a chapter on the data hunt for love. So I enlisted my reluctant wife into an experiment. We both signed up for Chemistry.com, and then waited (and waited and waited, as it turned out) for the algorithms to match us.

Now I get a press release about a new Web app called WotWentWrong. Taking a page from Customer Relationship Management, it gives people a customizable form to send to the people they went out with who.... just stopped calling. The idea is simple. Let's say you had a date or two, and you thought it went really well. And then the other person appears to vanish from the scene. If you're the nervy sort, you call, and you might pick up the phone or even knock on the door and experience an excruciating exchange full of "um, well...etc." More likely, you don't call or visit, and you're left worrying: What was about me that he/she didn't like???

photo: BigStock.com

Now, using WotWentWrong, failed dating gets a feedback system similar to what users of Amazon or Netflix have been known for years. Dating, in this view, is a consumer experience, and the customer--or shopper--must optimize his or her own "offering" in order to achieve the desired goals. Some of this could no doubt be useful. On the standard template, the person doing the snubbing is asked to weigh in on what went wrong. The choices:

1) You don't pay for dinner when we go out
2) You don't make enough time for me
3) Too much fighting
4) You are selfish
5) You text instead of calling

But there can also be positive feedback, such as:
1) You are attractive
2) You are insightful
3) You are enthusiastic

Here's what's weird about this for me. For a million years or so, we have developed the most sophisticed systems imaginable to "read" other humans. First, we have our animal awareness. We pick up odors, postures, eye movements, casual touching, all of the things that fall under the rubric of "chemistry." Then we have richly textured and nuanced language, something our most sophisticated supercomputers (like Watson) still struggle with. Human-to-human communication is deep, especially when the parties are face to face. This process is analog.

In much of our lives, we deal with big organizations that understand us largely by means of standardized forms and metrics. Employees in large companies suffer through annual performance reviews, in which they are placed into boxes that never seem to fit. The IRS sends more boxes to check. The Numerati take fairly primitive data and then put us into tribes (ie. The people who give five stars to Godfather II, Moonstruck and The Seventh Seal). Sometimes they can predict our desires. But they really don't know us as individuals. Compared to the miraculous complexity of analog communications, all of these formal systems are crude.

Yet we're growing used to them and sometimes find them less theatening. They're disembodied. Perhaps bad news hurts less when it comes as a set of symbols, and not a distracted glance, a bored tone, a note of sarcasm or a slap in the face. But we learn more and live far more richly when we brave the analog world, even when things go wrong.

1 comment [view]

Stuck in Seattle

January 23, 2012

I flew to Seattle last week wearing a watch that runs slow. The watch was the least of my problems. I caught the flight, and when I stepped off, six hours later, I turned on my phone and saw a message that the Big Data roundtable discussion I was going to, sponsored by the Markle Foundation, had been cancelled due to the coming storm.

So I had two days to kill in a Seattle that would soon be snowy and dysfunctional--so dysfunctional, as it turned out, that I my return flight would be cancelled, giving me four days in the city. What to do?

First, a word about slow-moving watches: They serve a function. If they tell you you're running late, you are. That's useful to know. (And you might be even later than they say.) On the other hand, don't trust them if they tell you you're early. (This reminds me of another limited-use technology: the bicycle rear-view mirror. If it shows a truck barreling up behind you, you can bank on it. But if it indicates that the coast is clear, don't trust it. Turn around and look.)

I had a lot of writing to do last week. So a couple of days in the Vintage Park Hotel, across the street from Seattle's wonderful public library, sounded just fine.

The view from my hotel room

One trouble. When it snows in Seattle, things shut down, including the library. This was unfortunate for me, but much worse for all the street people there who find not only warmth and comfort in the library, but also reading material and Web access. (Once when I worked at a newspaper in El Paso, I wrote about a Thanksgiving dinner at a soup kitchen. I sat next to a ranchhand-turned- hobo, who told me that the holidays were the worst of times, because libraries were closed for so many days. He was a big reader, as I recall.)

Outside Seattle's Library

A few other things I did in Seattle. I read a great book, John Vaillant's The Tiger. It changed the way I think about cats, human evolution, Russia, Perestroika, hunting, and many other things. The writing is wonderful, though I did find a few too many Russia names to digest. On a related note, I saw Werner Herzog's 3D documentary, The Cave of Forgotten Dreams. It takes you inside the fabulous Chauvet cave in southern France, where the art is 30,000 years old. It's worth seeing, even with Herzog's whimsical and awe-struck narration.

I also went to the Seattle Art Museum. A few items I liked there:

Caterpillar Suit, by Walter Oltmann, 2007

Detail from Paolo Uccello panel, from around 1450. (Did you know that he was also a mathematician?)

An ancient Greek who might look beautiful in the Caterpillar Suit.

2 comments [view]

The secret Jeopardy match: One year ago, Watson won

January 13, 2012

Jeopardy book

Last winter it snowed every week. And I remember tossing and turning in bed, worrying whether a storm might cause me to me the match between IBM's Jeopardy machine, Watson, and its human foes, Ken Jennings and Brad Rutter. If I missed the match, which took place a year ago this week, my book project would be highly compromised.

I was full of worries. There was the very real concern that Jeopardy executives, fearful that the results of the match might leak before it came out on TV a month later, would block me out. That possibility, it turned out, was very real, IBM sources told me later. Big Blue had to lean on Jeopardy to get me a seat at the performance, held at IBM Research in Yorktown Heights, NY. If it had been held at the Jeopardy studio, in Culver City, Calif., as originally plan, I have little doubt that I would have been blocked. As far as the Jeopardy team was concerned, I represented risk.

If I missed the show, the book would have been a mess. I had written 9/10 of Final Jeopardy, and it had gone through the editing process. But the last chapter hinged on the match. It was to take place on a Friday. I was to write the last chapter over the following two days and submit a draft on Monday. That would be edited and added to the book. A couple of weeks later, the public would be able to buy the partial ebook--everything except for the last chapter--online, and then would be sent the last chapter when the physical book came out, the day after the televised match. It was tight scheduling. For it to work, I had to get into the show.

My other fear was that Watson would lose. The machine lost about 30% of its matches against tournament of champion competitors in its last series of sparring matches. I had seen its vulnerabilities. Despite its strengths, entire categories could confound Watson. What's more, Jennings and Rutter were the best players on earth. Following the match, I've read lots of opinions on social media that IBM had fixed the match, and wouldn't have played it if there was a chance that Watson could lose. This is not true. And if Watson lost, my book would be the story of a machine that failed. Hardly a selling point.

I drove up to Yorktown from my home in NJ and didn't relax until I had gotten through security and was inside IBM Research. By that point, I figured that even if Jeopardy kept me out of the studio, I could watch on TV monitors in the overflow room. I wasn't particular. But I was concerned about capturing the data. I had a digital recorder, and for backup, I'd downloaded a recording app on my iPad. I would have to recreate the match, with the precise clues and scoring at each juncture, from the audio.

Much of the rest of that day I included in the final chapter: A wound up David Ferrucci, Watson's chief developer, crying as the make-up woman worked on his face; IBM CEO Sam Palmisano, as Watson steamrolled the humans, telling one of the researchers, "Maybe we should have toned it down a notch:" Jeopardy host Alex Trebek gamely entertaining the restive crowd with his stand-up routine during the seemingly endless technical glitches; an exhausted Jennings and Rutter, standing in a lonely hallway, waiting to do more video interviews after losing the marathon match, this while the happy IBM crowd was upstairs drinking cocktails and toasting the victor. Here are Jennings' memories.

As I drove home 52 weeks ago, I leaned out of the window and took one last picture of IBM Research on a winter night.

3 comments [view]