Real AI software team building...

From: ben goertzel (
Date: Tue Apr 02 2002 - 09:12:08 MST

Eli wrote:
> I understand that your opinions on this topic are strongly felt and have
> reasons underlying them. But the fact is, you have never led an
> all-in-one-place development team, *nor* a substantial
> development team. This doesn't mean your intuitions are wrong, of
> But it does give me less inclination to accept them.

Ben, why would I rely only on my own intuitions in a case where it's easy
just ask people? I've checked with people who have more software
development experience. I checked with *you*, remember? You said that
while you started Webmind Inc. believing in distributed development and had
people scattered all over the place, if you wanted to do it all over you'd
rather have had everyone in one place. If you've changed your mind since
then, I'll take that into consideration, but I will not actually change my
current plan, because all the *other* experienced software developers I've
talked to about this have said: "Yeah, if you want to take a serious shot
at this you'd better have full-time people on-location, it's not realistic

There is no doubt, the best case is having the best team all in the same

However, in real life one is always making compromises, and there are lots
of other factors.

I started out Webmind Inc. with a distributed team, and while this wasn't
optimal, the work that I did with this distributed team was instrumental in
helping me get funding -- funding that I used to build a less distributed

Even with funding, though, I was unable to find the right people in New
York, where I was located at the time. I found great AI scientists in New
York, but not great software engineers with deep AI knowledge. I found a
group of great software engineers with deep AI knowledge in Brazil, and I
am glad I decided to use them instead of just restricting myself to local
people. We are still working together now, a year after Webmind Inc.
dissolved. We have never worked together in one place, though we have
visited each other often, sometimes for up to 1-2 months.

So I've made all sorts of mistakes to do with geography & project
management! First there was the mistake of underestimating the value of
F2F communication, and relying on a wildly distributed team too long after
funding made it unnecessary. Then there was the mistake of *too badly*
wanting to work with people in New York, which led me to hire inferior
NY-based software engineers -- rather than accepting that the Brazilian
team I'd found was SO MUCH better than the New Yorkers I was able to
recruit at that time, that it was more effective to work with the
Brazilians at a distance with frequent visits. (This was 1998-99, a very
hard time to hire good engineers in the US; now the situation is a bit
different due to the changes in the economy. These days it might be a lot
easier to build a quality team in New York.)

So, obviously, the simple rules of thumb like "it's better to be colocated"
are just simple rules of thumb. Real decisions wind up being a lot more

In your case, my view is: If funding proves hard to come by, you shouldn't
overlook the possibility that working with a distributed team could help
bring your project to a point of maturity at which obtaining funding
becomes easier.

-- Ben G

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT