The news is under Watson’s hood, which signals a new era of intuitive computing and wide horizons for IBM [
NYSE: IBM]. The implications are far-reaching despite some misguided sniffs of derision from artificial intelligence cognoscenti, and are well beyond a single column. But let’s briefly consider two things; what it means for companies in IBM’s ecosystem, and what it implies about the emerging era of intuitive computing and The Cloud.
The last time IBM pulled off a similar stunt was several computer generations ago – the globally-watched 1997 chess match between Big Blue and chess master Gary Kasparov. Cynics suspected a PR motive on IBM’s part. A 2003 documentary film
Game Over: Kasparov and the Machine implied Deep Blue was an IBM plot to boost its stock. Even if true, why not? The stock did rise, for years, in the aftermath of that last stunt. IBM has an obligation to tell investors what they think the future looks like for their technology. Big Blue did the job. The world, and IBM, did change in big ways. Watson too was a pretty good way to tell the story of the next cycle.
(A Watson documentary is sure to emerge, but for now an excellent insider’s account of the team and stunt are to be found in Stephen Baker’s just-released
Final Jeopardy: Man vs. Machine and the Quest to Know Everything. And surely a Saturday Night Live Jeopardy spoof will emerge too, perhaps getting Darrell Hammond back to pit his
Sean Connery character against a computer played perhaps by the arm-flapping clanking “danger-Will-Robinson” robot of the 1960s TV show,
Lost In Space.)
IBM is one of the few still-standing companies from the Jurassic Era of computing with its roots starting, amazingly enough, before WWI and pre-dating even the first era of modern computing. There is no convention in how to denote eras in the history of computing, but the emergence of each great cycle has been associated with telegraphic names – and IBM has been there for all of them.
With the names Univac and Cronkite in 1952 we saw the launch of the first great cycle of the modern era of ubiquitous electronic computing — the era of mainframes made possible by the silicon transistor. It was presidential election night November 4, 1952 that legendary CBS newsman Walter Cronkite stood beside a UNIVAC that used statistics to predict the outcome. The traditional polls and experts heavily favored Adlai Stevenson over Dwight D. Eisenhower. UNIVAC said otherwise. Cronkite doubted, and postponed announcing the prediction. The subsequent rise of mainframes, in which IBM not UNIVAC proceeded to dominate, took IBM stock up over 400 percent over the following decade or so.
With the names Apple and Superbowl in 1984, we saw the second great cycle in computing – the era of the desktop made possible by the microprocessor. Apple’s [
NASDAQ: AAPL] unprecedentedly expensive $3 million (in today’s dollars) ‘Big Brother” 1984 Superbowl commercial heralded the launch of the Mac which epitomized, even if it did not dominate, affordable desktop computing. IBM took a few years to get their rhythm in that cycle to come up with a competitive desktop PC, and then dominated (for a while). IBM stock ran up another 400 percent or so in the decade or so following that era’s launch.
Now with the names Watson and Jeopardy in 2011 we see the beginning of the third great cycle – made possible by ubiquitous intuitive supercomputing parked in The Cloud. There is always symbiosis in these cycles. Network connectivity for example, the wired Internet, was a powerful accelerant for the desktop era. In this third cycle the unleashing of wireless broadband connectivity in handheld smart phones and smart tablets will be like proverbial fuel on the fire. But the real “smarts” are not in the phones and tablets – they’re in the Cloud, in staggeringly gargantuan arrays of Watson-like supercomputers. Your iPhone is only smart and fast enough to connect in real time to the Mother Ship of Smarts in the Cloud.
In each past cycle many companies flourished (not just IBM) and many died (not just UNIVAC). The same is about to happen again. Oh I know that Watson is not a “super” computer by modern supercomputer standards. Measured in flops (logic operations) the language of the digerati, the fastest supercomputer in the world is now China’s Tianhe-1A which set the current world record last year at 2.6 petaflops per second. (A peta is a thousand trillion.) That’s some 40 times faster than Watson. Keep in mind though that we are now living in a new stratosphere of computing power. Watson is over one thousand times faster than a 1980s state-of-the-art defense department supercomputer.
But flops are like horsepower. Once you have a lot of them you gotta know how to handle them. A lot of horsepower is a necessary but not sufficient condition to win a Nascar race. In computing, the driver that determines how the engine is used is the character and elegance of the algorithms, not only the programming logic itself, but also the skill in creating that logic in the first place. That’s what IBM properly brags about in the Watson project. Watson’s horsepower is controlled by IBM’s DeepQA operating software.
The difference between the conventional computing that has dominated since 1952 and DeepQA (and others similar) is the difference between mathematics and literature. The difference between understanding phrases or constructs that contain, in effect, words like one-plus-one, versus words like Paris Hilton. The former has only one answer, and is not context relevant. The latter has at least a couple of answers and the context matters a lot. The former, linear mathematical logic, is what dominates computing now; the latter requires intuitive or genetic algorithms. The latter requires considering, in parallel, more than one, often many possibilities and permutations, weighing the most likely in context, and picking an answer that is most appropriate not necessarily “correct,” or in many cases, about right.
The first two eras of computing, mainframe and desktop, were rooted in processors running blazingly fast mathematical software where rapid brute force calculations provided all that was needed for commerce and games. (Graphics rendering is inherently mathematical.) If you want a computer to answer natural language questions – essentially the Jeopardy goal (well, in reverse on Jeopardy) – it’s all about context, nuance, location, history, and a multitude of subtleties.
If you want to ask your iPhone, or the computer-generated voice at airline reservations, a normal question you’d sure like a normal answer. We’re not talking about simplistic already available voice recognition software where you can say something like “call Mom or “define petaflop” to your smart phone, and get a useful result. We’re talking about intuitive contextual recognition of what you mean when you ask a question like “what’s the next flight to New York?” and your phone — more properly, a Watson-type supercomputer your phone is seamlessly and nearly instantaneously connected to — knows where you are, where you’ve traveled recently, how you like to travel, your budget and airline preference and myriad other variables relating to your personal history and the world around, and then displays, or says, a number of options that make sense for you, now.
Watson can’t do that yet. And Cornell professor of science and technology Trevor Pinch, unimpressed, properly noted in his
column this non-trivial limitation in Watson’s skill set. Watson is not the epitome but the beginning of the next era of intuitive computing. Sitting by itself, stationary in a studio, Watson did well. Thrown in to the real world it would do less well with context-laden questions you might ask, far from home, about your flight delayed by storms.
Pinch cogently observed that it “is the informal, tacit, embodied knowledge that is the hardest for computers to grasp, but it is often such knowledge that is most crucial to our lives.” True of course. But does anyone doubt that given a few years of progress in horsepower and algorithm refinements, Watson won’t leap forward rapidly in capability? And we can’t ignore that much of the necessary embodied knowledge is now, and increasingly, available as you permit your smart phone to provide real-time GPS location information and link calendars, photos, web searches, purchasing behaviors and preferences to the Watson-filled Cloud. We are mere years, not decades, from getting intuitively useful answers from computers, in real time, that many of us will like.
The Watson-type arrays will populate in the Cloud in the thousands of huge buildings, data centers, already around the world, with more proliferating. And who benefits from all this? Ultimately we all do, and doubtless consumer-centric and hand-held based features, services, and games, will be the mega-drivers of demand. But first the benefits will accrue to high-end uses, the kinds of applications IBM itself is targeting in which intuitive computing will change the game as much as mathematical computing did for banks and computer games. Genetic algorithms and intuitive logic are what is needed in such areas as urban planning, and traffic control/prediction, in medical diagnostics, manufacturing, and even legal issues. Internet traffic itself will benefit from Watson-class software – not how you write emails, but how to manage the explosive traffic problem arising from video everywhere on the network.
All of this is not news to those on the inside of the compute community. It is well-studied and eagerly pursued, in part because of the sheer excitement of the technical challenges as author Baker illuminates regarding the IBM team, and in part because it is bound to be a very profitable business.
The business ecosystem Watson epitomizes ranges from the chip and server makers, to Cloud data center builders and operators. It’s a monstrously large ecosystem today, and growing rapidly. It’s about to grow a lot faster as we ramp-up this, the third great era of computing. Whether IBM experiences another 400 percent decade-long rise in its stock value may be too much to expect in this cycle, but it’s not crazy.
There are lots of others well-known, not so known, and emerging who will be propelled by this emerging trend, just as there were in the last two great cycles. It is a fertile domain for serious investors to plow. Some likely and obvious beneficiaries are certainly IBM competitors like HP [
NYSE: HPQ], Dell [
NASDAQ: DELL], and Fujitsu [
NASDAQ: FJTSY].Cisco[
NASDAQ: CSCO] believes the datacenter boom benefits them, as does Juniper [
NYSE: JNPR] with the latter’s
clever new architecture – architecture that recognizes the data center as, in effect, really just a massive new type of supercomputer. Of course, Google [
NASDAQ: GOOG] gets all this and is arguably a pioneer as well.
And then there are the builders and operators of the Clouds (data centers) themselves, the likes of Rackspace [
NYSE: RAX], Akamai [
NASDAQ: AKAM], Equinix [
NASDAQ: EQIX], DuPont Fabros,
[NYSE: DFT], and Savvis [
NASDAQ: SVVS]. All this explains why Verizon [
NYSE: VZ] purchased data center behemoth Terramark [
NASDAQ: TMRK] last month for $1.4 billion in cash, and why Bezo’s is bullish on Amazon’s [
NASDAQ: AMZN] Elastic Compute Cloud service. Depending on what they have up their sleeve, perhaps there is growth salvation in this great cycle for Microsoft [
NASDAQ: MSFT] too. But just as Microsoft’s future was invisible to IBM at the dawn of the second cycle, both these players surely know there are similar relatively tiny players today running below the radar, poised to innovate and grab great gobs of market share in this third era of computing.
The names Watson & Jeopardy will go down in history alongside similar previous iconic names that dramatized great pivots. Baker, author of the aforementioned
Final Jeopardy,
said “Machines like Watson are going to become part of our lives.” To be sure. And there will be tens of thousands, indeed likely millions of the more powerful progeny of Watson. But most of the action, unlike the IBM PR stunt, will take place invisibly in The Cloud. The Third Era begins. <>
http://blogs.forbes.com/markpmills/2011/02/21/ibms-watson-jeopardy-stunt-unleashes-a-third-great-cycle-in-computing/
