Tuesday, October 12, 2010

ORF 2010 - Day 1 - Monday


What a great day. Lots of fun and good food at lunch. Here's a play-by-play recap of some things that I noted today. Anyone who feels maligned, well, feel free to drop by and we'll share a Glen Livet and see why I say that.

Carole Ann Berlioz-Matignon: (She's dropped the Berlioz these days and I kind of miss it.) She said, (1) "Don't expect business users to think like developers." Well, OK, but they actually can. And, not only that, they kind of HAVE to think like developers and developers have to learn how to think like business users. (2) "Business guys need Tables, Nets and Graphs." Again, not true - nobody HAS to think like anything because we're all different and we all think different so we can't assume that all business users think like that. QED. (3) "Again, focus on the Business Concepts and NOT the rules." Ouch. The rules ARE the business logic!" Business Concepts and rules are one and the same thing. BUT, the presentation was simply glorious - she's using some kind of really cool software that allows her to give really cool, really neat presentations. BTW, since she made probably 150 documented points and we only disagreed on three, that's pretty close.

Pete Voss: (1) "Animals do not do abstract thought." Actually, animals have been shown to be able to count, communicate and express tremendous feelings for each other and ever some for humans. I just have to give animals more credit. (2) "Neural Gas Architecture." I don't know what that is yet but I'll be working on Goole overtime tonight. Supposedly it's a Dynamic Architecture based on data. Sounds really cool. BTW, way back in the dark ages of Ge transistors and diodes we (the good old USAF) did VERY large-scale training projects. Sure, the computers took up several buildings, but we did them.

John Laird: Now this guy was at CMU with The Forgenator in the early day of AI and Rulebased systems; a true rule guru. He still insists that people need to remember where we started and the this is, after all, AI. Dammit! (OK, he didn't say, "Dammit" but should have.) He claims to be the first person to write a rulebase (a game that he invented in college) with over 1,000 rules.

Rolando Hernandez: Rolo pretty much showed WHY we need to keep knowledge from "leaking" out of the company but some pretty fair ideas on how to to do it.

George Williamson: Something about "Forecasting performance of games." Didn't quite catch all of that but maybe I'll pick it up on the slides later. (2) He commented about taking a line-by-line translation of a C program straight to Java and the problems that they had with it. Well, Duuuhhh.. NEVER translate a program line for line. Not even C to C++. (3) "Business Users cannot understand complex coded rules." Horse hockey! I've worked with business users who did it. They did it because they wanted to do and (maybe) because their jobs were on the line. (4) "You can't test all the rules." Again, well... I did that at Lloyds Bank because we HAD to have 100% verification because it was an on-line banking problem. 800 rules had 65K+ test cases but it had 100% verification. (5) "RBS is a nightmare to maintain from an IT point of view." NO! RBS helps relieve the nightmare if the business guys are writing and testing the rules, NOT the IT guys. Once you put the monkey on their backs, they will make sure that everything works and works right! I guess that traveling around and not staying on one job more than 18 months gives you a different view. (OK, now that's just bragging and promoting traveling consultants, isn't it??) But, I have to admire a guy who stays on for 12 years at one place and puts up with the constant internal politics and wins out - that's something I could never do. :-) Finally, (6) there was a comment about "not having the same time on each device in the USA" (mostly in Texas, I think) Again, back in 65-67 we (NASA) kept everything synched up while doing satellite tracking at lunar distances using WWV out of Ft. Collins, CO, located at 5, 10, 15... MHZ. Nothing more than time hack every second of 1KHz and maybe 10 cycles that we zeroed in every shift. I don't think I'm going to get a job with Union Pacific after this but it had to be said.

James Owen: (moi) Forecasting with Rules. OK, lots of chit chat and didn't hole the putt with the first stroke and the finish was anti-climatic. (Forgot a slide showing how the C code of a forecasting rulebase done in 1992 would look in OPSJ or Drools in 2010. So sue me!) And I was really tired and missed completing the job. So, now I need to go back and put all in the stuff that I left out. Maybe Jason will give me another chance for redemption next year.

Luke Voss: Missed it. One of my favorite speakers (just me him last year) and I had to go get new keys from the front desk to change shoes. Oh, well. Maybe he'll tell me about it later and explain the slides.

Panel Discussion. Well, I deserved that since I started it at ORF 2009 last year. I was kind of hoping that Jason wouldn't do that this year but we're doing it. And the attendees seem to like it. A couple of photos later.


More tomorrow! Tune in.

No comments: