back to article IBM's Watson-as-a-cloud: Is it a bird? Is it a plane? No, it's another mainframe

IBM's attempt to spin its supercomputer-cum-TV star Watson into a $1bn business unit may eventually boost Big Blue's bottom line – but going from beating Jeopardy! to defeating cancer is going to be harder than expected. The system's decision engine and advanced natural-language processing technology was launched as a new …

COMMENTS

This topic is closed for new posts.
  1. alain williams Silver badge

    IBM should ask Watson itself

    If it is as good as it is cracked up to be it will be able to tell its keepers how best to use it to earn $10bn per year :-)

    1. Anonymous Coward
      Anonymous Coward

      Re: IBM should ask Watson itself

      I'm not sure if you're trying to be funny, if so disregard from here.... Watson doesn't work like that, it's not creative in such a way, what it does is give you answers to direct questions with %age confidence. It can't build and design things.

  2. Anonymous Coward
    Anonymous Coward

    for some organisations a tool like this could be useful

    Name one?

    That's a serious question, beyond stock trading companies (who already have purpose designed systems for doing this), I can't think of a single use for this (albeit very interesting and innovative) technolgy.

    1. Suricou Raven

      Most obvious? Customer services. Watson makes a decent conversational agent. A little work and it could take over the role of first-line support, quite capable of dealing with the most common customer queries. A little less frustrating than the telephone maze game, and potentially cheaper than hiring a building full of telephone monkeys.

      1. Anonymous Coward
        Anonymous Coward

        Most obvious? Customer services

        Yeah, that's going to work.

        I already won't phone some companies at all, because doing so is pointless. All it is possible to get is what the company decides their computer should be capable of dealing with, which normally covers only a small amount of the reasons why someone might phone them.

        Using computers to replace humans, whilst cheap, is a sure fire way to really, really piss off, and lose customers.

    2. Anonymous Coward
      Anonymous Coward

      It's already being used by healthcase in the US to suggest methods of treatment for particular combinations of conditions, these are then - obviously - ok'd by humans.

      Watson could be baked into all of IBM's software, instead of static help pages, Watson could say "it looks like you're trying to setup inter site replication for your TSM servers. Do you want me to make suggestions as to how you might optimally configure your networks, based on my knowledge of your site and TSM?

      1. John Riddoch
        Thumb Down

        "it looks like you're trying to setup inter site replication"

        Just what we need - Clippy on steroids....

        1. Anonymous Coward
          Anonymous Coward

          Just what we need - Clippy on steroids....

          Yes, but remember when we had operating systems from some two-bit, jumped up little company (Microsoft I think they were called), and these were crap and kept crashing?

          Then along came Linux and sorted itself out from the chaff. So it could be cool.

      2. Anonymous Coward
        Anonymous Coward

        Watson could be baked into all of IBM's software, instead of static help pages

        And all you have to do is get your management to pay for the vast amounts of storage required for the database Watson requires to be able to suggest how you can do the job they're paying you for.

        1. Anonymous Coward
          Anonymous Coward

          You wouldn't have a Watson on each site, the software would connect to one back at IBM, just like normal online expert systems, I would have thought that was obvious. That said, looking at your handle, maybe you're just trying to be contrary for the sake of it.

      3. ld0614

        Or in lotus notes it could ask, It looks like your writing a letter, would you like some help with that?

        1. Anonymous Coward
          Anonymous Coward

          Or in lotus notes it could ask...

          Ah, looks like you are using Lotus Notes. Would you like me to find some other, less shit, programs you could use instead?

      4. CRConrad
        FAIL

        Where's the icon for...

        ...for Clippy the paperclip when you need it?

        (Fail, because the Reg failed by not providing a Clippy icon)

    3. Anonymous Coward
      Anonymous Coward

      Having said that... I suppose it would be a useful technology for the analysis of telephony and internet metadata...

      "Watson - Does Mr Joe Average communicate with anyone who isn't American?"

      "Watson - Has Mr Joe Average ever visited a website containing any anti American sentiment?"

      Maybe the Chinese would like to buy a copy to run on their new 'quantum computer'?

      1. Destroy All Monsters Silver badge
        Paris Hilton

        Why so negative?

        I'm sure if there is a good use for it, it will be found. Remember the "whole world needs 5 computers" stuff? Your imagination may well be less developed than it needs to be.

        Here is a good use: politicial scoring: Hey watson, I have this letter here reported on the Daily Beast, "Nuclear Weapon Free Iran Act of 2013"; here are the endorsers. Any of those faggots somehow related to PNAC? ... "Yes, Dave, they are all neocons and several have close ties to AIPAC, AEI and FDD."

        And "the Chinese" have a quantum computer?

        1. Anonymous Coward
          Anonymous Coward

          Here is a good use: politicial scoring...

          Except we don't need a computer to do that, we have aides, "What would be a good way to piss off the mayor... close two lanes of the major bridge route into the city".

    4. Ethangar

      Biggest one that jumps to mind would be medical diagnostic. There is no doctor on earth that can keep up with all the research being done the world over even if its in a very narrow field. Have watson gather and collate all the information published, research papers, findings, studies etc so that your doctor could just call and tell watson that patient x is presenting with x,y,z symptoms, these drugs didn't do anything what is another course of treatment? Patient has XX what are the new treatments? etc. I think something like that would be quite useful.

    5. Ian Michael Gumby

      @obnoxious Git

      This would be useful in fraud detection. But there are already custom solutions in that space.

  3. Anonymous Coward
    Anonymous Coward

    IBM is going nowhere with Watson

    Three interrelated reasons:

    - Resources: the pool of domain experts at IBM is shrinking or no longer there at all. After many years of layoff after layoff they don't have any resident domain experts able to establish a sensible conversation with its potential customers. And throwing hundreds of off-shore resources -however cheap they are- at a knowledge based project simply does not work.

    - Risk: IBM has become so risk averse that nowadays does not want to invest in anything. When they approach customers with new ideas they always start talking about "sharing the risk" with their customers. Translation: you'll pay tons of money to IBM for doing this and that money will be used by IBM to build knowledge that can be reused to repeat the project, possibly with your own competitors. IBM will not risk a cent should the project fail, they have fantastic teams of lawyers (which together with finance are their core competence now) that will make sure this happens.

    - Technology: IBM is playing catch up in all fronts. It has not developed by itself any new software or hardware technology in the last what, ten years? They act like CA or Oracle, gobbling up new technologies and "integrating" them into more expensive deals and milk from existing customers their expensive support fees, but not gaining any significant new business from technology. They may have a huge patent portfolio providing revenue, but 1- as these patents expire there are no new ones replacing them and 2- the cost of developing such a portfolio cannot be recovered by patent fees alone, they'll have to deploy the new technology, but for the first two reasons (resources and risks) that will not happen. Unless the company changes dramatically, of course.

    I believe IBM will be in the future a case study on how short term profit goals can destroy an unprecedented amount of brand image and trust in a short record time.

    1. Anonymous Coward
      Anonymous Coward

      Re: IBM is going nowhere with Watson

      Complete rubbish.

      - "Domain experts", I've never heard the expression before, but let me assure you IBM does not want for experts in its systems.

      - Risk: IBM pumped something like €50M into the Munich Linux programme, that was nothing if not risky.

      - No-one else has anything like Watson, which pretty much disproves that comment.

      1. Anonymous Coward
        Anonymous Coward

        Re: IBM is going nowhere with Watson

        IBM is a great example of some very serious long term research investment (you might differ on what). A company that I know well who have completely abandoned fundamental applied research is Hewlett Packard. Now that really is an example of a company who have sold out to quarterly results.

      2. Anonymous Coward
        Anonymous Coward

        Re: IBM is going nowhere with Watson

        <<- "Domain experts", I've never heard the expression before>>

        That may explain your lack of understanding and uninformed dismissal of the comment? AKA "subject matter experts" these are the people that have knowledge of a market or business, often because they have been involved for years in projects in these sectors and have developed close relationships with customers. They are usually middle level analysts or managers, and are the next target of layoffs in developed markets/countries after all the low level people have been off shored. IBM still has a few of them, but no intentions of hiring more or replacing them when they retire or are "resource actioned"

        <<- Risk: IBM pumped something like €50M into the Munich Linux programme, that was nothing if not risky.>>

        That project being risky depends on how you see it. Barring Munich going bankrupt or disappearing, it is unlikely that they were not going to recover that (see the section on lawyers) It is still far shot from 1 billion, if you ask me.

        <<No-one else has anything like Watson, which pretty much disproves that comment.>>

        Perhaps because no one else sees any use of it, IBM can't find anyone that wants it, especially at the cost they are asking for it? And can you mention anything else where IBM is leading in technology?

        1. Anonymous Coward
          Anonymous Coward

          Re: IBM is going nowhere with Watson

          Are you seriously suggesting that IBM don't have subject matter experts? It's such a wrong comment that I don't even know where to begin.

          And no, in 16 years working in IT, many of them at as a subject matter expert (in Backup/Storage) I have never heard the expression "Domain Experts". I have regularly heard the term Subject Matter Expert though.

          1. Anonymous Coward
            Anonymous Coward

            Re: IBM is going nowhere with Watson

            <<Are you seriously suggesting that IBM don't have subject matter experts? It's such a wrong comment that I don't even know where to begin.>>

            No, what I'm suggesting is that IBM is getting rid of subject matter experts as time and pressure to keep the share price growing (resource actions) and they have no intention of replacing them.

            <<And no, in 16 years working in IT, many of them at as a subject matter expert (in Backup/Storage) I have never heard the expression "Domain Experts". I have regularly heard the term Subject Matter Expert though.>>

            You need to step out of your niche a bit more, I guess.

          2. CRConrad

            Re: IBM is going nowhere with Watson

            That means your experience is too narrow, not that expression is new or unknown. Let Google do a count for you, or something, and you'll see.

        2. oldcoder

          Re: IBM is going nowhere with Watson

          Microelectronics/nanoelectronics.

          solid state physics

          Holographic storage.

          nanofabrication

          Supercomputers

          The very disk drives you are using were created by IBM technology.

          1. Anonymous Coward
            Anonymous Coward

            Re: IBM is going nowhere with Watson

            <<The very disk drives you are using were created by IBM technology.>>

            Which supports my point. That was 20-30 years ago. Patents from back then are expiring, with nothing replacing them neither as revenue source from the patent itself or from implementation of the technology.

            <<Microelectronics/nanoelectronics, solid state physics, Holographic storage, nanofabrication>>

            My point was (agreed by others), none of these technologies is being implemented by IBM. This is because their risk aversion combined with their focus on next quarter results.

            <<Supercomputers>>

            So IBM are somehow leading the pack in that area? How?

            1. CRConrad
              Facepalm

              You seem to be the wrong person to be dissing Big Blue.

              <<The very disk drives you are using were created by IBM technology.>>

              >>Which supports my point. That was 20-30 years ago.<<

              No -- it was forty, fifty, perhaps closer to sixty years ago. What are you, born in the nineties or something, to have such a remarkably shallow perspective? History didn't begin yesterday, nor even the day before yesterday.

      3. Ian Michael Gumby
        Boffin

        @AC Re: IBM is going nowhere with Watson

        "- No-one else has anything like Watson, which pretty much disproves that comment."

        Actually they do.

        Many companies run on COTS technology. Some are even trying to define it. (Face Book)

        If you read the article Watson is based on a couple of Apache technologies. (Hadoop and UMIA)

        Companies who are going to invest in Watson have found it cheaper and having less risk by doing it themselves with hired consultants who have the domain expertise, not only in the tech, but also in their business domain too.

        1. Anonymous Coward
          Anonymous Coward

          Re: @AC IBM is going nowhere with Watson

          I think you fundamentally mis-understand Watson, if you think it's just a few smartypants database queries run against hadoop and umia. There is no COTS version of Watson, it's not just some sort of expert system.

    2. Ian Michael Gumby

      @AC (OP) Re: IBM is going nowhere with Watson

      I guess the other AC was the one who down voted you.

      While you are both ACs, I can wager one is a current IBMer and you escaped the Borg.

      I can attest to your post that you're pretty much spot on.

      Although I will disagree on the point that IBM hasn't invented anything. They have, but they have not been able to take it from the labs to production where it can be monetized.

      There is obviously more that I can say, and like you, I've escaped the borg too... ;-)

      1. Anonymous Coward
        Anonymous Coward

        Re: @AC (OP) IBM is going nowhere with Watson

        I'm the OP. And no, I've not escaped the borg. I just had to deal with it during more than half a decade of "partnership" and some of my colleagues where TUPEd (or local equivalent) to IBM. So I know a bit about how it works.

        And, yes, the one downvoting and not hearing the term "domain expert" is likely someone still working for IBM and quite happy with its storage architect role.

  4. Quentin North

    The ultimate jeopardy answer to life the universe and everything.

    42.

    Now, Watson, what is the question?

    1. Nick Ryan Silver badge

      Re: The ultimate jeopardy answer to life the universe and everything.

      "Now, Watson, what is the question?"

      What is seven times eight?

      1. PM.
        Facepalm

        Re: The ultimate jeopardy answer to life the universe and everything.

        seven times eight is 56.

        That was easy.

        1. Graham 24

          Re: The ultimate jeopardy answer to life the universe and everything.

          The question is "What do you get if you multiply six by nine" ?

          I had to scrabble about it to find the question, but it's definitely the question. It gives the right answer. Of course, it helps if you work in base 13.

  5. Destroy All Monsters Silver badge
    Paris Hilton

    It's better than MYCIN for sure, but...

    Rather, for IBM's cloud-based service to generate insights greater than the sum of an individual developer's contributed data

    But that's not the goal of this system, right? It should be an "adviser system", so in the classical example of medicine, you have this set of information about a patient and you ask the watson what the hell is up with the patient; it will follow the text links through its medical dictionaries and papers and come up with a few hits. It's search. Will it be able to discover that the patient has two different problems? Possibly not. Can you bolt a probabilistic reasoning engine onto the side? Probably not. Should Doug Lenat be gotten out of retirement? Maybe yes.

  6. Stevie

    Bah!

    This story is an obvious fabrication.

    The mainframe has been obsolete since the mid-90s as any early 90s poll will show you and therefore could not be housing this "Watson" as IBM would have thrown them all away by 1999.

    Moreover, even if it *were* on a mainframe everyone knows it would have been deleted by management as you weren't allowed to have gaming software on them because it slowed them down or something (never really understood the pointy-hair arguments myself).

    1. Ian Michael Gumby
      Boffin

      Re: Bah!

      The mainframe hasn't been "obsolete" but too expensive to run and maintain.

      The problems solved on the mainframe can still be solved on a mainframe, but can also be solved cheaper on a Linux cluster. (Note that the mainframe can also run Linux within an LPAR)

      The big disadvantage of a mainframe is that you can't tear it down and reuse the components for another project if necessary.

      1. Stevie

        Re: Bah!

        If the application is so mainframe-worthy, why in Seymour Cray's name would I be "tearing it down" for parts?

        Sounds like these :Linux" servers you cite have an in-built obsolescence cost somewhere, possibly one measured in years as opposed to decades.

        And if you thing processor virtualization is a child of the toy-computer technology, you obviously haven't really worked in the parts of the industry you are discussing.

        Remember: If you can lift it without a crane, it isn't a computer, it's a toy. Don't run your mission critical software on toy computers (with apologies to Optimum Cable TV for stealing the meme).

  7. Peter 39

    no problem

    All this commercial/Jeopardy stuff is just a front for the NSA-oriented systems. Who knows how many they've bought (tell us in the notes if you know :)

    But if you need deep pockets, these are your friends.

  8. southen bastard

    the answer is out there

    The best use for watson is an interconnected dateing site, that takes the data you supplied, cross refferances face book, tweeter, tumbler than checks all other dating sites finds that you told a heap of lies, " you are not a 20 somthing, athletic, non smoker with 15inches hanging between your legs" however finds your perfect match anyway, just insert your credit card details here, then you will be able to go away and bred a whole swag of credit cards and I can have too.

This topic is closed for new posts.

Other stories you might like