Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.1 6/24/83; site opus.UUCP Path: utzoo!watmath!clyde!burl!ulysses!mhuxl!ihnp4!zehntel!hplabs!hao!cires!nbires!opus!rcd From: rcd@opus.UUCP (Bob Busy) Newsgroups: net.flame Subject: Re: interesting news article - AT&T AI projects Message-ID: <653@opus.UUCP> Date: Tue, 31-Jul-84 19:07:22 EDT Article-I.D.: opus.653 Posted: Tue Jul 31 19:07:22 1984 Date-Received: Sat, 4-Aug-84 01:05:56 EDT References: <292@ihlpf.UUCP> <1174@vax2.fluke.UUCP> Organization: NBI, Boulder Lines: 65 From moriarty, who was commenting on AT&T projects intended to increase net traffic... >For the sake of posterity, I think we should mention several of the earlier, >more pioneering projects: > > 'the rich rosen / dick dunn parallel abuse device' > This was one of the more famous of the early "random > name-calling" devices, where a 'rich rosen' process was > created which gainsaid everything its prototype stated; the > process ID for this complement program was 'dick dunn' > (notice the common use of first name/last name identical > initials). Both were programmed to escalate their co-abuse > at an exponential rate; a great deal of new ground was > broken by the programmers, who had to delve deeply into > medical textbooks to provide enough data for the processes' > constant derogatory references to the other's sexual > limitations. 'One of our most exciting experiments', said > one enthusiastic researcher. Moriarty seems to be a little confused on some of the details. Having been involved with one part of this project (the so-called "dunn" process), let me add a few interesting notes. The network identity is even sillier in the larger context. "Dunn" was created as a person who had supposedly done graduate work in computer science under William Waite - note again the first/last identical initials, not to mention the sophomoric humor in choice of last names. The rosen and dunn processes were an experiment in extending operating system theory into non-cooperating, semi-communicating processes. In fact, we were surprised at the extent to which these processes would cooperate, considering that they were developed independently, with the extent of collaboration being that of ensuring that they would receive each others' submissions. They were run from locations widely separated on the net (Boulder and Piscataway), such that propagation delays could be used to allow charges and countercharges to arrive in indeterminate sequence. This aspect was marginally successful; in the end it seems to have confused some readers so much that they didn't bother to reply, and the net result appears to have been less traffic rather than more. Moriarty is confused in particular with the class and targets of insults. In fact, the dunn and rosen processes regularly attacked the decvax!murf but very seldom each other. This, incidentally, produced some interesting behavior from "murf", since the dunn and rosen attacks often targeted the exact same statement but for completely opposing reasons. We have some reason to believe that "murf" ended up trying to execute obscure portions of code which had not been debugged, as the responses of the murf process eventually started to lose their coherence. Also, the references to sexual processes were confined to the rosen process - that was a concept which we simply hadn't considered. We might attempt to add it in some future version. Speaking of future work, we're now considering several new directions. I'll just mention one in passing: The rosen process generates copious output directed to net.music, including fairly frequent references to the "minimalist" music of Cage. We originally opted to do very little in music, simply to allow us to focus elsewhere. However, we did make a provision in the dunn process for occasional mischievous notes about the Grateful Dead, on the chance that this would eventually provoke a response which could start a real battle. We intend to allow the next release of the dunn process to make a connection between Cage and the Grateful Dead; we expect that any resulting submissions should provoke at least the rosen process and probably some of the real readers as well.