David Anderson

From BOINC Wiki
Jump to: navigation, search

1. Why the name BOINC? Did you have a couple of options to go with before choosing this acronym? If so, what were the others?

I wanted something light and non-techie.
I got the idea of the sound a drop of water makes falling into a pool: "boink".
I sort of reverse-engineered the acronym (Berkeley Open Infrastructure for Network Computing) but I later got rid of this because it's cumbersome.

2. Did (anyone from) Berkeley have anything to do with the naming, did they want Berkeley mentioned?


3. How did you get into contact with the first co-developers? Did you advertise something, or did they offer themselves to you? Or did they move over from Seti Classic?

I wrote the guts of it (client, scheduler, web) by myself in about 6 months in early 2002.
Then I got a grant from NSF, and hired Charlie Fenton, who had worked on [email protected]
I hired some UC undergrads during the summers.
One of these, Seth Cooper, wrote the first GUI (which was integrated with the client, not a separate program).
I started getting major volunteer contributors, including Janus Kristensen, Bruce Allen, and Carl Christensen.
Later Rom Walton started working as a volunteer, and I quickly hired him.

4. How did you feel when you first saw BOINC being officially used by a million computers?

The biggest thrill was pre-BOINC, early in [email protected] when we got 1,000 years of CPU time in 24 hours.
The biggest thrill in BOINC was when great projects like Einstein, CPDN etc. started using it, as well as unexpected projects like [email protected]

It was also a big thrill when people like Matt Blumberg, James Drews and Willy de Zutter developed significant companion systems like accounting sites and account managers.

5. Did you reach all your goals set with BOINC? Or is there still something missing that you hope will be added?

My goals aren't even close to being reached!!
I want volunteer computing to be recognized by the HPC world as a peer of cloud and cluster computing.
I want there to be conferences about volunteer computing, and papers published, and PhD degrees granted.
I want thousands of scientists to benefit from volunteer computing, not just a few dozen.
I want everyone in the world to know about the existence of volunteer computing;
I want 20 million desktop/laptop computers and 50 million phones running BOINC.

6. Of you could do it all again, would you change anything? If so, what? (Can be in the client, the forums, the back-end, the personnel)

The main things would be:
- Concentrate on a simple, game-ified user interface from the start.
The BOINC Manager interface is too techie and drives off non-techie people.
- Create a centralized mechanism for allocating computing power, rather than the "project ecosystem" model.
The problem with the latter is that a new project cannot be guaranteed any computing power at all, so creating a new project is too big a risk for scientists.

7. With BOINC being a governance, has something really changed in the past year - besides Rom leaving us?

Of course. All the tasks that Rom used to perform are now being divided among other people, including volunteers and part-time people like Christian.
Many software contributions are being integrated in the source code without my ever seeing them.
We're getting a lot more contributions than before.
The main idea is that BOINC should survive even if its funding gets completely cut off, or if key people unexpectedly die.

8. Is there anything you secretly hope someone would start a project about, something you may have been toying with for some time now?

I'd like to see more evolutionary systems.
I think there was one for walking gaits for multi-legged robots;
I'd like to see more ambitious AI projects, or human-guided evolutionary art or music projects.

I don't necessarily want "the singularity" (when AI's can increase their own intelligence) to happen, but volunteer computing would be useful tool toward that end.

9. What's next on the agenda for BOINC? Would developing the back-end to work on a server run with GPUs be something?

What's next on the agenda is what I'm currently working on:
- a framework in which volunteers sign up for science goals rather than projects, and their computing power is allocated by a central mechanism.
- adding BOINC-based back ends to major conventional HPC centers (Texas Advanced Computing Center, nanoHub, and hopefully many others).

Also on the agenda:
- GPU computing on smartphones
- move to VM/Docker as the standard way of deploying apps; vastly simplify the process of creating BOINC projects.

10. To recognize you as a true human, what's your favorite game, book, food, sport (aside from climbing), vehicle, number, colour, animal?
game: cryptic crossword puzzles
activities with my son Noah (11): Kerbal Space Program, model rocketry.
book: "White Noise" by Don deLillo
food: India Pale Ale (well, that's actually a drink)
sport: I've competed in: distance running, triple jump, volleyball, basketball, table tennis
I do for fun: tennis, ultimate frisbee, badminton (the fast/competitive kind)
vehicle: Piaggio LT150 scooter (my commute vehicle)
number: the continuum c: the cardinality of the real numbers
colour: earth tones
animal: cats
music: late romantic and early 20th-century classical