One fine day at meshU

May 21, 2008 ⋅ 2 Comments »

I just got back from meshU, a one-day conference focused on design and development for the web. I went on a bit of a whim; the student tickets were only 30 bucks, and there were lots of interesting speakers. Well, it was a good decision — the three talks I saw were well worth the price.

(As an aside, I see that I’ve been added to Patrick Mueller’s Planet OTI aggregator. OTIers, read on! Today’s post contains at least two OTI connections.)

Avi Bryant: Turning the Tables: Moving Beyond Relational Storage

The first talk I attended was by Avi Bryant of Dabble DB fame. Avi talked about why how not to use a RDBMS. There’s been a lot of interest lately in alternatives to the relational database — CouchDB, Amazon SimpleDB, and Google BigTable being the most famous examples.

Avi mentioned that there are two main cases where it’s advantageous to have something other than a relational database. First, if your dataset is massive, as with Google. In this case, it’s just not feasible to use a RDMS. The second case is if your dataset is really just many small, independent datasets, in which case it might be simpler and more scalable to use an alternative technique.

At the end of the talk, Avi demoed MagLev, which is a Ruby interpreter built on top of a Smalltalk VM (OTI connection #1). This was really cool, not just because it’s Ruby running on top of Smalltalk. It was cool because MagLev can transparently persist heap objects to disk, and distribute them to other VMs, either locally or running on other machines. For example, you can define a function in one VM, and then call it from another VM. Wow. As far as I know this was the first time MagLev has been shown in public, but there’s going to be a talk at RailsConf 2008.

Daniel Burka: Iterative Design Strategies

The second talk I attended was by Daniel Burka (OTIers probably know Daniel’s brother Peter). Daniel is the lead designer for Digg and a co-founder of Pownce. Daniel gave a great talk on iterative design strategies. You can check out the slides here.

It was really interesting to hear some of Daniel’s stories from Digg. He said that one of the times that they redesigned the comments, they were getting a lot of negative feedback from users. But they also noticed that under the new system, stories were getting more comments than ever before. So although there was a vocal minority who weren’t happy with the design, it was an overall success. The moral of the story is that you need to look at the implicit feedback as much as the explicit feedback.

Another thing I took away from Daniel’s talk was that I should read How Buildings Learn by Stewart Brand. I’ve had this recommended to me enough times that I’ve finally decided to buy it.

John Resig: Building Interactive Prototypes with jQuery

John Resig is the creator of jQuery, “the write less, do more Javascript library”. This talk was pretty timely for me. I just started using jQuery a few weeks ago for a Firefox extension that I’m writing, and I’m pretty much in love with it. So it was practically guaranteed that I would like this talk, since it was further demonstration of all that is awesome about jQuery. I did learn one new thing from the talk — I’d never heard of the jQuery Form Plugin before, and it is really cool as well.

I’m not sure what more I can say about this one. If you’re a web developer, or a designer who codes a little, you should seriously check jQuery out.

---

So that was my day. $30 well spent at meshU. Thanks to the organizers: Mark, Mathew, Rob, and Stuart.