[TOS] Bridging research and implementation in open source

[David Humphrey]

> Part of the strategy behind this is trying to create a buffer between  
> the very energized industry folks who want to help and the very  
> resistant higher education culture.  At least at our school,  efforts  
> by industry, however well intentioned, are seen as intruding on the  
> educational process.
> 
> If you have suggestions at how we might best approach this, I'm all  
> ears!

We've been discussing this in the Mozilla context for a bit now, and I 
have some thoughts, again inspired by the fact that I think we want to 
leverage what we are/have in order to build what we need.  Here are the 
facts as I see them:

1) Many profs/researchers are doing specific work on X, and aren't 
particularly interested in applied work--they write papers instead of 
code (this isn't universally true, I know, but it is the general case).

2) Many projects (I'll speak for Mozilla) are working on hard research 
type problems (JIT, static analysis, cache architecture, HCI) 
specifically from an implementation/engineering point of view.  That is, 
they often don't write papers but code.

3) Many profs/students are looking for project work in open source for 
their students.

I have observed that these three groups often remain separate.  I know 
open source developers who are working on things you'd think researchers 
would care about, but can't attract attention.  I know researchers who 
are doing work that should find its way into open source projects, but 
doesn't.  I know profs/students who work on toy projects instead of real 
work.

Now I leave the realm of fact for an idea :)  What if we help bridge the 
gap between these worlds, and connect them?  What if we help to create 
partnerships where we take on the "engineering exercise" that flows from 
primary research?

Let's say you're a researcher with interesting ideas about cache 
architecture, and over here is Mozilla with a vehicle for your work to 
reach hundreds of millions of users *and* over here is a group of 
students in India and some more in Toronto and Denmark who want to work 
on something real.  These groups currently can't work together very easily:

* The researcher (typically) doesn't want to do implementation.  It's 
hard work (hard in a different way than the paper she just wrote), and 
won't necessarily further her career at the university.  But, she knows 
all the theory and could explain it to someone else.

* The open source project would be interested in the project, but 
doesn't have the resources to try the experiment.  They do have the 
knowledge of how you'd do it though, and could advise people.

* The prof/students studying software engineering or doing a capstone 
project don't know the researcher or the open source project, and 
certainly can't connect those dots on their own.  But they have energy 
and the need to write some code (i.e., they are being marked).

What's needed here is a body that knows all 3 groups, and can provide 
the connective tissue to make this partnership happen.  From experience 
doing partnerships between the 2nd (open source) and 3rd (students) 
groups, I know that this can work.  I've never tried with all three, though.

This idea would depend on collaboration between a large and diverse 
group: you need people who know the researcher and can speak her 
language; people who know the project and can determine if this has 
value and would get accepted; people who know how to get answers to 
implementation questions within the project; people who could work with 
students to do the actual work.

I'm someone who is capable of working at the bottom of that stack, but 
I'm not a CS researcher.  I know there are those on this list who are.

What do you think?

Dave