By Sramana Mitra and guest authors Shaloo Shalini and Bhavana Sharma
SM: Right, which is where startups play very well. Startups start with the lead users and begin developing the market, and once they gain momentum and have validated the space, you pick them up and deploy them to a broader population.
DF: So, I think going from automation to optimization is one area where we are going to push hard. Another one is probably around something in knowledge management. A lot of what happens with IT management software happens because there is a big gap between the people who use the IT systems and the people who deliver them, so there is a big gulf between the data center and the people who are using it. A lot of that gulf is due to the lack of the ability to deliver information. I think a simple example of this is the common service desk request, is a problem. I am trying to do something and it doesn’t work. The vast majority of the time, what happens is you make a phone call to the person and say you type this, go does that. What is happening is that the service desk analysts are actually functioning like an information broker. They are basically getting requests and figuring out what the answer is.
SM: But this Web self-service solutions have existed for a while. I remember there was this company called Kanisa that was doing that back in 1999–2000.
DF: Yes. But the thing that seems to be missing is twofold – one is that you have a vocabulary problem. The example that I give is, I on a PC had a Windows service IPS Core Service (IPSSVC), which was going nuts and has taken over the machine. I did a Google search to figure out how to fix it, but I know enough to know that the search I wanted to submit was “Windows Service run amok,” not “IPSSVC run amok” because I know that IPSSVC is a Windows service. There may be a general solution that one of the people on my tech team is working on. This is the thing that’s missing – the shared ontology.
SM: Exactly, and in fact if I remember – you may want to look into it – Kanisa was acquired by several companies and today it exists as Knova. It has that entire infrastructure for setting the taxonomies and ontology and is a Web self-service solution that can automatically configure itself. Of course we have to set it up. Once you set it up, I think it provides that kind of capability.
DF: Right, but I see there is a white space in the sense that nobody is delivering such a thing in a general way.
SM: Yes, I think the gap you are pointing out, I figure what is going on that product, as far as I remember, was primarily brought into the customer service organization and not to the IT organization and you are looking for that technology as it applies to an IT organization.
DF: In fact, the other aspect of it is kind of spooky, but we are piloting some of it is which is it also can be a learning system. Which is where you do not need to set the topology up, it can learn. It’s kind of knowledge extraction, so it is a way of building tribal knowledge.
SM: It’s a very complex artificial intelligence (AI) problem, though.
DF: Yes, it’s one of these things. It’s kind of funny that you say that. When I was in grad school I was in software systems and mathematical performance, and there was a AI team of professors, and of course all those guys who would be doing spooky math would look down on the AI people. We used to have a saying: AI people tell you they want to go to the moon. Then they climb an apple tree and tell you they are closer to their goal. What I would say is yes, that’s true, but they also have apples now. I think that solving the problem completely is a very complicated, almost impossible task, but I think you can do some relatively simple things that would have a significant impact.
SM: That I agree with completely. You know, I have done learning products before in my career, and I think paradigms are very powerful. More and more it’s a type of technology that I think the world is more ready now for because there is so much data on which to apply those learning algorithms.
DF: Part of it is data; part of it is the fact that we can actually observe communication. Data at rest is interesting, but what you actually want to do observe the messages, and a lot of message exchange now is done in an observable form: e-mail, instant messaging, comments on Web pages, and so forth, so you can watch the exchange and the way kids learn. Kids learn by watching messages that are exchanged in a form they can understand. We can now exchange messages in a way that a computer can watch. I think it’s an opportunity to do some basic learning to get all ontology that bridges the gap between IT and the users of IT.
SM: Very interesting. Thank you for your time, Don, and I enjoyed the conversation very much.
DF: Yes, I enjoyed it very much, too, and I think about these conversations are kind of triggers and changes in thinking, so I find them very helpful. Thanks again.