Eye on the i World: What You Think You Know About Watson Might Be Wrong

Analytics & Cognitive - Other
  • Smaller Small Medium Big Bigger
  • Default Helvetica Segoe Georgia Times


What's Watson? That high-end IBM analytics platform that only a global health insurance company or some other giant conglomerate can use effectively? Maybe not.

I'm starting to be a bit mystified by the IBM Watson phenomenon, and for more than one reason. On one hand, there's at least a segment of ISVs in the IBM i market that think Watson's nothing to be concerned about. On the other, there's IBM itself, which seems to be hoping IBM Watson will be a big engine of growth in its corporate future, and yet, some of the information IBM offers about the platform via the web seems, well, surprisingly vague and even unintentionally misleading.

The Lack of Concern Among Some IBM i ISVs

There appears to be a fairly nonchalant attitude toward Watson on the part of some IBM i software vendors I've talked to in recent months. Mostly, they seem think that there's no threat to business intelligence activity on the IBM i and that possible convergence between the two platforms could be a long way off.

For example, last March I interviewed LANSA's Steve Gapp in "Eye on the i World: Complacency and User Aging Threaten the IBM i," and his comment on Watson was, "I don't think IBM i software vendors will face a threat from Watson specifically, but they will face threats by not keeping pace with current trends in Web and mobile spaces."

Last May, I interviewed Rocket Software's Dan Magid in "Eye on the i World: The i as the Center of a Multiplatform Ecosystem," who said, "At Rocket, we continue to look at IBM i and its core applications as the center of a multiplatform ecosystem. We are investing heavily in the IBM i space with a focus on providing the infrastructure that allows our customers to stay out in front of the latest technology trends." I should note that one of Rocket's products is Rocket Discover for Cognos TM1. (IBM Cognos TM1 is a product that IBM offers as an adjunct to IBM Watson Analytics.)

Last June, I interviewed Profound Logic's Alex Roytman in "Eye on the i World: Software Tool Vendors Sees IBM's Watson as IBM i Ally Rather Than a Threat," where Roytman stated, "IBM i is typically the 'system of record' that uses traditional relational databases. This will not be replaced by Watson, but rather integrated with it where necessary."

Last July, in "Eye on the i World: BI Tool Vendor Sees Positive Perception of IBM i as Key to the Platform's Future," I interviewed NewGeneration Software's Bill Langston, who offered, "For most IBM i customers, IBM's Watson technology will slowly come into the enterprise through interfaces built into future releases of their business application software. It's going to take a few years for this to develop, though, because at this point I don't think many customers or ISVs feel confident that they understand how to get started and how this technology might be profitably applied to their business or industry." I should note that NGS offers NGS-IQ, a leading IBM i business intelligence application suite, among other data-analysis products.

Last month, when I approached another IBM i business intelligence software vendor, which I won't identify because it declined to do a Watson-oriented interview, a vendor representative explained the company's demur on the partial grounds that "none of our customers is interested in Watson."

Granted, this is not a statistically valid sample of vendor sentiment, but even anecdotally these comments signal an attitude that many ISVs may share. While such comments could be interpreted as admirable self-confidence on the part of vendors who have been pillars of the IBM i community for decades, taken together, they can give a different impression.

How can it be that a platform, the traditional main strength of which is the handling and analyzing of databases to facilitate business performance, has no room on its radar for IBM Watson, which can analyze larger volumes and more types of data sources for exactly the same reason? Or are those who think Watson's nothing special with regards to the IBM i (not necessarily including the people I've just quoted) simply whistling past the graveyard?

IBM's Watson in More Than One Kind of Cloud

Setting that aside for the moment, let's take a quick look at IBM's current marketing of Watson on the web so we can scratch our heads a bit more.

Let's start with IBM's most public web page for Watson, "Build Your Cognitive Business with IBM." When I went there, I wanted to answer some questions that seem pretty basic to me:

            •What's the cutoff point for the amount of data a user needs to analyze that might form a sort of boundary between enterprises that could profitably use Watson and those that can still get by with DB2 and SQL or third-party tools?

            •What's an estimate of the size of a commercial enterprise that could benefit economically from the data analysis available from a Watson purchase?

            •What will the impact of Watson via the cloud have on IBM i vendors that sell business intelligence and reporting software for in-house use or that make their money offering consulting services to companies who need help with packaging their internal data so non-techie execs can understand it?

What I found were brief rundowns on Watson products, which were helpful in getting the Big Picture but didn't answer my questions. I found a lot of discussion about the value of "cognitive computing" that were largely a new spin on "know your market," a concept even Ebenezer Scrooge's father must have been familiar with. I found discussion of the need to explore "dark data," which turned out to be mostly data from documents such as memos and e-mail messages. Gathering that information in an automated way would be helpful, no doubt, but as to how to do it, the main answer seems to mostly be "Watson can do it, trust us." Rather opaque for someone asking, "How can I use Watson in my case?"

Elsewhere, I found a video touting case studies about big companies getting useful results. Legends Hospitality, a national tavern chain, predicted sales by analyzing past customer data. Paschall Truck Lines consolidated hiring and retention of its nationwide truck driver cadre. Caliber Patient Care optimized the routes its ambulances take. Mears Group corralled the 600 million records that make up the worker injury reports they handle.

Nice examples to be sure, but all are big companies, not SMBs. In fact, one gets the distinct feeling that in the context of such enterprises, SMBs are out of their league here. How much data was analyzed? Except for Mears, no mention. How small could a company be and benefit from Watson? Apparently, not too small at all. What will the impact be on IBM i vendors and consultants in the IBM i data analytics market segment? "Insufficient data, Captain," as Mr. Spock might say.

On this page, IBM also self-poses the question, "Why do I need it [Watson]?" and its answers are "to answer business questions," "to take confident action on insights," "to tell a compelling story," and "to analyze trusted data." Gee, I'd say enterprises have been trying to do that at least since the invention of the AS/400 and more like since the close of World War II. It's not too self-evident what Watson's value might actually be, and those questions don't link to other pages. If I were an SMB looking at such web sites cursorily, I might easily conclude that there's nothing there for me.

If you dig, though, you'll find that there are some cloud-based Watson services that might be affordable and highly useful, if lacking in descriptive detail. But do you have a couple of hours to watch all the videos and browse all the web pages IBM provides? If you're like most people, not really, unless you already have a mandate to embrace the Watson platform. So you move on, and Watson becomes another concept you don't have time to learn to understand. This makes another question emerge: Why isn't IBM doing more to emphasize Watson's usefulness to SMBs? Because clearly, there is some.

Debunking Myself

One could plausibly argue that Watson is just for big companies that need to analyze Big Data. Conventional thinking might support that. After all, Watson Analytics, for example, runs on high-end Power Systems servers like the 750. Pricey! Nothing for SMBs with limited IT budgets here, right? No wonder some IBM i ISVs aren't concerned. Except…

Watson Virtual Agent, which helps automate customer service departments, runs on Mac or PC, requires only a browser, and costs $265 a month per subscription. Watson Explorer, a cloud-based SaaS package that analyzes whatever structured and unstructured data you care to give it and includes an application builder, just needs a server running AIX, Linux, or Windows. Watson Analytics, a data analysis and visualization service that uses a natural language interface rather than SQL, is accessible as a cloud service and costs $30 a month per user. There's even a freeware version you can try out for nothing! (Wouldn't that beat learning SQL yourself or waiting three weeks for your overburdened IT department to come up with four properly structured queries you'd really like to run now?) Watson Knowledge Studio can be trained to mine unstructured text (like e-mails and memos) for useful data and includes annotators for specific industries. That runs $150 a month for one subscriber.

These prices are not out of reach for most SMBs, particularly if the information gleaned ends up paying for itself with expanded economic opportunities for the enterprise. The products are available as cloud services, so there's none of the muss or fuss of licensing and maintaining software in-house if you don't want to go that route. I won't do more than mention the Watson APIs available that could interface with your existing software to improve analysis and reporting.

So does it seem like some IBM i ISVs aren't worried about Watson? Maybe they should be. Does it seem like IBM itself is focused too much on marketing Watson mostly to giant companies? Maybe that's a mistake. Do you think Watson has nothing to offer you if you're an IBM i SMB? Maybe you should rethink that.

Over the next months, I'm going to try here to resolve some of the questions we all may have about Watson. What is Watson, and what can it do? What does it mean to the future of the IBM i market? How can you determine if it's something in your near future instead of way over the horizon somewhere? What's the future for companies that consult on data analytics if these tools are directly available to SMBs via the cloud, eliminating so much need for someone in the middle? The answers to these questions will be significant. We could all stand to get beyond the Jeopardy! championship hype and see if Watson might mean more than we currently think it does.

Let's figure this out together.