This has been a whirlwind week at Magenic. Monday night Rocky Lhotka (the ‘h’ is silent and the ‘o’ is short, by the way), a Magenic evangelist and Microsoft Regional Director came into town and presented at the Atlanta .NET User’s Group. The developers at Magenic got a chance to throw questions at him before the event, and then got another chance to hear him at the User’s Group. The ANUG presentation was extremely good, first of all because Rocky didn’t use his power point presentation as a crutch, but rather had a complete presentation in his head for which the MPP slides — and later the code snippets (see Coding is not a Spectator Sport) — were merely illustrative. Second, in talking about how to put together an N-Layer application, he gave a great ten year history of application development, and the ways in which we continue to subscribe to architectures that were intended to solve yesterday’s problems. He spoke about N-Layer applications instead of N-Tier (even though it doesn’t roll off the tongue nearly as well) in order to emphasize the point that, unlike the in COM+ days, these do not always mean the same thing, and in fact all the gains we used to ascribe to N-Tier application — and the reason we always tried so hard to get it onto our resumes — can in fact be accomplished with a logical, rather than a physical, decoupling of the application stack. One part of the talk was actually interactive, as we tried to remember why we used to try to implement solutions using Microsoft Transaction Server and later COM+. It was because that was the only way we used to be able to get an application to scale, since the DBMS in a client-server app could typically only serve about 30 concurrent users. But back then, of course, we were running SQL Server 7 (after some polling, we audience agreed it was actually SQL Server 6.5 back in ’97) on a box with a Pentium 4 (after some polling, the audience concluded that it was a Pentium II) with 2 Gigs (it turned out to be 1 Gig or less) of RAM. In ten years time, the hardware and software for databases have improved dramatically, and so the case for an N-tier architecture (as opposed to an N-Layer architecture) in which we use two different servers in order to access data simply is not there any more. This is one example of how we continue to build applications to solve yesterday’s problems.
The reason we do this, of course, is that the technology is moving too fast to keep up with. As developers, we generally work with rules of thumb — which we then turn around and call best practices — the reasons for which are unclear to us, or simply become forgotten. Rocky is remarkable in being able to recall that history — and perhaps even for thinking of it as something worth remembering — and so is able to provide an interesting perspective on our tiger ride. But of course it is only going to get worse.
This is the premise of Vernor Vinge’s concept of The Singularity. Based loosely on Moore’s Law, Vinge (pronounced Vin-Jee) proposed that our ability to predict future technologies will collapse over time, so that if we could, say, predict technological innovation ten years in the future 50 years ago, today our prescience only extends some five years into the future. We are working, then, toward some moment in which our ability to predict technological progress will be so short that it is practically nothing. We won’t be able to tell what comes next. This will turn out to be a sort of secular chilianism in which AI’s happen, nanotechnology becomes commonplace, and many of the other plot devices of science fiction (excluding the ones that turn out to be physically impossible) become realities. The Singularity is essentially progress on speed.
There was some good chatting around after the user group and I got a chance to ask Jim Wooley his opinion of LINQ to SQL vs Entity Framework vs Astoria, and he gave some eyebrow raising answers which I can’t actually blog about because of various NDA’s with Microsoft and my word to Jim that I wouldn’t breath a word of what I’d heard (actually, I’m just trying to sound important — I actually can’t remember exactly what he told me, except that it was very interesting and came down to that old saw, ‘all politics are local’).
Tuesday was the Microsoft 2008 Launch Event, subtitled "Heroes Happen Here" (who comes up with this stuff?). I worked the Magenic kiosk, which turned out (from everything I heard) to be much more interesting that the talks. I got a chance to meet with lots of developers and found out what people are building ‘out there’. Turner Broadcasting just released a major internal app called Traffic, and is in the midst of implementing standards for WCF across the company. Matria Healthcare is looking at putting in an SOA infrastructure for their healthcare line of products. CCF – White Wolf, soon to be simply World of Darkness, apparently has the world’s largest SQL Server cluster with 120 blades servicing their Eve Online customers, and is preparing to release a new web site sometime next year for the World of Darkness line, with the possibility of using Silverlight in their storefront application. In short, lots of people are doing lots of cool things. I also finally got the chance to meet Bill Ryan at the launch, and he was as cool and as technically competent as I had imagined.
Tuesday night Rocky presented on WPF and Silverlight at the monthly Magenic tech night. As far as I know, these are Magenic only events, which is a shame because lots of interesting and blunt questions get asked — due in some part to the free flowing beer. Afterwards we stayed up playing Rock Band on the XBOX 360. Realizing that we didn’t particularly want to do the first set of songs, various people fired up their browsers to find the cheat codes so we could unlock the advanced songs, and we finished the night with Rocky singing Rush and Metallica. In all fairness, no one can really do Tom Sawyer justice. Rocky’s rendition of Enter Sandman, on the other hand, was uncanny.
Wednesday was a sales call with Rocky in the morning, though pretty much I felt like a third wheel and spent my time drinking the client’s coffee and listening to Rocky explain things slightly over my head, followed by technical interviews at the Magenic offices. Basically a full week, and I’m only now getting back to my Silverlight in Seven Days experiment — will I ever finish it?
Before he left, I did get a chance to ask Rocky the question I’ve always wanted to ask him. By chance I’ve run into lots of smart people over the years — in philosophy, in government, and in technology — and I always work my way up to asking them this one thing. Typically they ignore me or they change the subject. Rocky was kind enough to let me complete my question at least, so I did. I asked him if he thought I should make my retirement plans around the prospect of The Singularity. Sadly he laughed, but at least he laughed heartily, and we moved on to talking about the works of Louis L’Amour.