The conference was opened with a talk by Brian Ripley. I’ll distort his talk into 3 points that came across to me.
1. R Core is finite
The time available from R Core members is a strictly limited good. The more that is pushed onto R Core, the less attention to details. That doesn’t sound good. The more things that can be pushed out of R Core, the more attention to details. Sounds better to me.
For those of us not in R Core, the use of R is a privilege. There is a metaprivilege that goes with the privilege: the privilege of using R is not going to go away on a time scale that any of us needs to worry about.
3. Packages rule
R Core granted us three wishes. I’m not sure how to phrase the first two wishes, but the third wish was more wishes. The package system allows us to add whatever functionality we want very easily.
Other people have summarized Brian’s talk more literally. These include:
- David Smith from Revolution
- Colin Gillespie at Why?
- Karl Broman on Statistics, genetics, programming, academics
- Pairach Piboonrungroj on Tourism Supply Chains & Logistics
Have I missed any?
There were three talks tied for second, but the one I saw was a brilliant complement to the first. Brian’s talk gave us insight into the center of R. David’s talk showed us the vast ecosystem that surrounds that center. The visual portion of the talk is worth a look.
The new kid on the block regarding finance is R-adamant. There is some overlap with Rmetrics but it is coming at things with a different approach. It just came on-line — what is available is a small fraction of what is planned.
At the conference I had two surprises regarding “The R Inferno”. My first surprise was the large number of people at the conference who didn’t know of it. A group of people who are motivated enough to come to a conference dedicated to R would seem to be a group that would diligently seek out documents on R.
In one session I asked for a show of hands of those who hadn’t known of the Inferno before the conference. About a third of the room raised their hands. (Meeting people was closer to random sampling but a smaller sample, and suggested a higher number.) Next came my second surprise: I asked those who did know of it if they were surprised that so many raised their hands. No hands went up.
My surprise is obviously some combination of being too close to the Inferno and not knowing how bad searching for R documents already is.
This isn’t about one document — no matter how amusing it might be — but the general issue of getting the most useful documents into the hands of each R user. There was talk about problems as the number of packages approaches 10,000. But the number of documents about R is a few orders of magnitude greater than the number of packages. We’ll be talking millions of documents before long.
Connecting users and documents is going to get harder and harder. We need to find ways to increase the probability of good matches.
The last session of talks on the first day was an hour of lightning talks. Each speaker had 15 slides. There were 3 slides between speakers. All slides appear for precisely 20 seconds. One quick question was allowed during the changeover.
Lightning talks are absolutely wonderful for the audience. They were the talk of the town.
However, lightning talks are a tough medium for speakers. To be effective:
- slides need to be very visual with few words
- the speaker must practice, practice, practice
Lightning talks are not only good in their own right, they are good training for longer talks.
The final official event was Frank Harrell describing plans for useR!2012. It will be 2012 June 12-15 at Vanderbilt University in Nashville Tennessee.
My final event happened to be a spontaneous chat with Frank and others about ideas for next year’s conference. If you have ideas, Frank is keen to hear them.