First Day of School


Today is my (Bob Carpenter‘s) first day as a research scientist at Columbia University Department of Statistics. I’m part of a decent-sized team working for Andrew Gelman on implementing scalable Bayesian inference.

Having spent almost thirty years on the school calendar the first time around (from kindergarten in ’68 to tenured professor in ’96), the rhythm of the school calendar feels very natural. Some of my more senior friends have expressed envy at my (relative) lack of responsibility (for designing curricula, recruiting students and hiring faculty, raising funds, teaching, advising students, etc.). Ironically, when I left academia the first time, many non-academics thought I’d finally have to “work for a living”. In fact, industry’s a cakewalk compared to life as part of the professoriate.

Why Academia?

My motivation for moving is to have more time to spend learning about and working on hierarchical models of the kind I’ve been exploring for data coding and RNA sequencing. We’ll also be working on hierarchical models for social science, such as extending the multilevel logistic regression Andrew and crew used for Red State, Blue State, Rich State, Poor State to larger data sets, more predictors, more interactions, and more levels (the details of the basic setup are in Andrew and Jennifer Hill’s regression book).

What about the LingPipe Blog?

I’ll still blog here about natural language processing in general and LingPipe in particular.

I’ll move my general statistical blogging to Andrew’s blog.

I don’t know what I’m going to do with the bioinformatics stuff, but I may just start another blog for that. If you haven’t guessed, I think blogs are most useful if they’re focused.

What about LingPipe and Alias-i?

I’m still working part time for Breck at LingPipe*. I’ll still be answering e-mail and supporting the software. I’ll even be adding to it as needed.

We’re about to roll out version 4.0.1, which fixes some inconsistencies in the tutorials and fixes some non-standard definitions and some bugs in the scored precision recall evaluation. It’ll also have blocked logistic regression.

I’m also about to roll out the next draft of the LingPipe book. It’s now 350+ pages, with lots more general and LingPipe-specific content.


*Breck’s changed the name of the company from “Alias-i” to “LingPipe”. It’s déjà vu — my last employer changed its name from “Applied Language Technologies” to “SpeechWorks” to match the product name. Breck’s working on the rebranding and he’s hired our friend Philip Riley to work out the design side of things.

5 Responses to “First Day of School”

  1. Jeff Dalton Says:

    Congratulations! Welcome to the dark side ;-).

    Good luck in the new job. Hopefully, we’ll get a chance to meet at conferences or get you to come up and give a talk at UMass.

    We’re doing some work here in the CIIR and in the new computational social science research cluster that you might find relevant.

  2. Rich W Says:


    Congrats on the move – but definitely don’t stop blogging. I’ve really learned a lot reading your posts and I hope you can be as prolific on Andrew’s blog as you have been on LingPipe’s. Best of luck!


  3. Emre Sevinç Says:

    Congratulations and good luck. I hope you’ll still have time to work on the LingPipe book.

  4. Kathy Baker Says:


  5. Dave Lewis Says:

    Good news for Bayesian hierarchical modeling! Have fun.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: