A touching story

Gestalt is an idea a friend of mine and I came up with towards the end of the summer of 1996. Basically, it's a way of using Java to distribute parallelizable computations over a large number of processors. After the initial conception of the idea, we talked it over with another smart guy I know, and the three of us decided that the idea was good enough to pursue. So we did. We spent the last few weeks of our summer investigating the economics of getting people to pay for time on the virtual machine created by the network, and of paying people to have their idle computers donate cycles to the network. We also considered paying people for their donations by connecting them to the Internet for free. We thought we were going to make a lot of money.

And so we went and talked with a guy whose job it is to give money to people with ideas that he thinks will make a lot of money. We were disappointed to find out that he didn't think ours would make any. In retrospect, I think he's right... Quite obviously so, in fact (which is why I'm a Computer Science major, and not Economics). What the hell were we thinking? Although we seemed to have looked at things from the worst case scenerio, our judgement was clearly influenced by our excitement in hoping that we might be in charge of something that had a lot of potential. But, duh, how would we make enough money to be able to pay for donator Internet connections? And where would we find people to use the network...and pay us? I think Mr. Dallal (the man with the money) saw very clearly the problem with our money making scheme, and stated it just as well: our product was a vitamin, not a painkiller. People would use it because it did something good for them...but they wouldn't mow pedestrians down with their car, squeel into a parking lot at 70 MPH, run to the counter of a store (leaving their keys in the ignition, the car running, and the door open), and scream at the top of their lungs WOULD SOMEONE PLEASE HELP ME FIND SOME GESTALT?!

So the three of us disbanded for the summer and went our separate ways to college. The other two eventually lost interest in the thing, but I, fascinated with our having gotten the chance to talk to someone like Mr. Dallal, continued to pursue it. I sent him some email, and he sent me some back, and eventually I found some free time and forced myself to crank out a limping, ugly version of Gestalt. I called it version 1.0, and ran some tests on the machines at school, thinking that the slowness of Java would kill any residual interest I had in the idea. Unfortunately, the results weren't bad. My slaves were churning out MFlops of computations, which resulted in a few GFlops for a network of a machines. I had convinced myself that the idea was worth pursuing. So I did.

I wrote another, somewhat better version of Gestalt, and called it 1.0 again. This time, I thought about the code a lot more carefully, and tried to come up with abstracts for the operations I wanted the program to perform. I didn't actually get the whole thing working this time around, because the end of the semester rolled around, and I had to study for finals and write a bunch of programs. I wrote a paper on Gestalt for my Networks class, and gave a little presentation on it. I decided that I didn't want to finish the version I had been working on, because there was a better way to do it.

Which brings us pretty much up to date, with me sitting at my computer, finishing Gestalt 1.0 (rewrite 3). I no longer think it's a thing that will ever make any money. Besides, something like Gestalt shouldn't be available only to those that can pay for it. I guess I have grown up in the past year or so, and now I just want to set it free. After all, isn't that the best way to be? So enjoy it! I think it's a pretty neat thing, and it would be silly to claim that I was the first to have thought of it. Instead, and all I want is a little credit for taking the time to code the Java (which I enjoyed anyway). Well, that and control over how the source files are distributed: I intend on making an official release soon, and I don't want crappy betas floating around the net, so please don't distribute these files if you download them.


Last Modified: 7/24/97 by jack@cs.hmc.edu