HN Explorer

Ask HN: What should I ask Dr. David Ferrucci, the leader of IBM's Watson project

4 points by georgi0u ↗ HN
A little background: I am a senior CS major at Rensselaer Polytechnic Institute, and this upcoming week my school will be holding a 3 day event to cover the Watson computer system taking on Jeopardy. The lead researcher, David Ferrucci, is a graduate of RPI and will be here one of those days to do a Q&A, but I don't really know much about the system and figured HN might have a few insightful questions that I could forward on.

Also, kind of a weird twist, but Ferrucci happens to be my roommate's first uncle (he's a management major and never mentioned it since him and I don't really talk CS all that often), so I might be meeting Dr. F personally afterward - making the question asking even more feasible.

4 comments

[ 3.3 ms ] story [ 25.0 ms ] thread
solipsist
Ask him what he thinks the various applications of the Watson project are (if he can disclose some of them). I heard they were going to use it for medical purposes (let Watson answer people's questions and help diagnose them), but I'm sure they have a lot more in mind. It'd also be interesting to have him compare this feat to Deep Blue.

Let us know the question you end up asking him and his answer, if you can! Thanks!

dandrews
You're his nephew's roommate? Ask him "May I send you my resumé?"

Seriously, I presume that his budget was large, but not inexhaustible. Ask him what he could have accomplished if he'd thrown more hardware/people at the problem.

georgi0u
I'm already employed, but when my roommate told me this you better believe I went off on him (in jest) over not telling me before, in the past 3 years of my knowing him.

That being said, I think the latter part of your comment this could be a good question. I think asking him what issues he ran into - which might be more appropriate after he explains what the particulars of the project were/are - might be a little more generic though.

As an aside, after I posted this I went and did a little research on IBM's site - which is more marketing fluff than technical explanation - and one of the things they emphasized was that Watson was built as a custom solution to overcome some (ambiguous) lack of technical feasibility with existing hardware. One of the questions that came to mind after hearing this was why they couldn't just use the same algorithm/ of a blue gene/X. Basically, what differentiates Watson from any other distributed super computer, if anything (significant). I imagine theres no big difference, probably just tweaks in networking types, cpu counts/speeds, etc. - things that would've made Watson doable, but not optimally doable.

kenjackson
Could you ask him what is the single most important thing that makes Watson possible, whereas Watson would not be possible five years ago? Was there an advancement in querying unstructured data? Or was it simply disks got big enough to hold a lot of data?