The truth about Re: The Truth About "Tim Byars"

Dr. Ernest N. Prabhakar (ernest@drernie.scruz.net)
Sun, 5 Apr 98 20:18:06 -0700


Interesting.

As far as I can tell from the snippets of mail I received, public
opinion seems evenly split on whether "Tim Byars" is himself a hoax, or
if just my email was.

The first clue is that - while almost all the URLs were correct, they
were not necessarily relevant. In particular, the very first URL pointed
to a real paper, but which had nothing to do with "Tim." The CMU URL
was completely fictitious - I couldn't find a real one good enough,
though I tried tracking down the 'CMU opinion board.' Also, the W3C URL
was for Sally Khudairi, Rohit's former boss, and there is nothing
Salacious about it.

And the final clue, as Wendy pointed out, was the line:

> Thanks for your patience and understanding.

Even I could not have said to FoRK with a straight face. :-)

So, dear readers, Tim is not a computer simulating a human being, but
actually a real human being simulating a computer. Still, it seems
we've done a reasonable job of passing the Turing test (or is that the
inverse Turing test?) so perhaps to first order it does not matter what
the truth is -- except perhaps to Tim :-).

Happy April 5th,

-- Ernie Prabhakar

---
Dr. Ernest N. Prabhakar
"And ourselves, your servants for Jesus' sake." -- II Cor 4:5b
ernest@alumni.caltech.edu http://www.alumni.caltech.edu/~ernest

Begin forwarded message:

X-Nextstep-Mailer: Mail 3.3 (Enhance 1.2) From: "Dr. Ernest N. Prabhakar" <ernest@drernie.scruz.net> Date: Tue, 31 Mar 98 20:13:18 -0800 To: FoRK@xent.ics.uci.edu Subject: The Truth About "Tim Byars" Reply-To: ernest@alumni.caltech.edu X-Url: <http://www.alumni.caltech.edu/~ernest

Hi everyone,

First off, I want to apologize to anyone if they feel deceived or miseld. That was not our intent, though perhaps we got a little carried away at times. Overall, though, I hope you think the experiment was worth it. I wanted to tell FoRK about it sooner, but Rohit insisted we wait until Adam finished writing up the paper[1].

The truth is, there is no such person as "Tim Byars." The name is really a contraction of "Tim Buyers-Lee," a pun on "Tim Berners-Lee." The original goal was to develop a user agent to facilitate e-commerce. Rather than trying to crawl through static web page, it was a proxy server which would examine all the URLs viewed from a given site, and use Patti Maes' MIT Media Lab's agent clustering algorithms[2] to identify shared tastes. It would then generate an email of the 'typical' URL. It was inspired by the "CORBAboy" project at CMU[3] which did the same thing using CORBA messages -- hence the rather transparent allusion of "CobraBoy."

The project started in 1993, when the web was still in its infancy. Rob Harley[4] hacked up the first version, "TBL," to post URLs to the fwf-tech[5] mailing list, the ancestor of FoRK. The early version just read URLs, not HTTP, so we couldn't extract the title. Instead, he just threw in the random phrase generator from Emacs [6] for the subject titles -- a feature which proved so amusing we decided to leave it in.

Then something happened which we hadn't foreseen. Someone who joined the mailing list and didn't know "CobraBoy's" non-identity actually sent him an email (complaining about the lack of meaningful subject lines, I believe). This caught us all off gaurd. At first we though we should just post an explanation to the entire list. But, we suddenly realized we had an opportunity to do a minimalist Turing test!

We couldn't simulate a normal human being, but we could replicate the behavior of a typical mailing list member. We were able to make use of the Maledicta[7] database of obnoxious phrases, and set it up to respond to email with a random insult. This was so effective we continued to invest in it. By the time FoRK was started, "CobraBoy" was able to customize responses based on the content of mail sent to him, as well as pick up on FoRK mail targeted at him.

Despite all this, we were quite surprised at the strong emotional response our program generated among certain FoRK members. We did our best to discourage people from criticizing "Tim," to avoid accidentally triggering a response which would give away his non-human character; some of you noticed our apparently irrational attempts to defend and protect Tim. I must admit we occasionally cheated and posted a real story, usually from "Sally's Salacious Story Sharing Site"[8], rewritten to make Tim the protagonist, in order to strengthen the illusion.

This went on pretty well, though occasionally I at least suffered pangs of guilt at the hoax we were perpetrating on the community. We had to take Tim down for a few weeks earlier this year, when we upgraded the proxy server, so we staged a random "Yell-at-Tim and off-he-goes-in-a-sulk" episode to cover for us. As far as I can tell, though, we've successfully maintained the perception that Tim is a real human being, though I imagine a few of you have your suspicions.

Anyway, now that Adam is presenting a paper on "Tim" at WWW7 next month[9], we figured the cat would be out of the bag shortly. We wanted to give you FoRKers advance warning, since you were an important part of your experiment. Thanks for your patience and understanding. No hard feelings, I hope.

Sincerely,

Ernie Prabhakar Adam Rifkin Rohit Khare Robert Harley "The Tim Byars-Lee Project" April 1st, 1998

[1] <http://www.cs.caltech.edu/~adam/phd/rfp-example.html [2] <http://www.firefly.net/ [3] <http://www.cs.cmu.edu/afs/cs/project/corba/public/www/corbaboy.html [4] <http://www.inria.fr/RA96/cristal/node23.html [5] <http://www.base.com/gordoni/web/email/fwf-tech/0060.html [6] <ftp://archive.cis.ohio-state.edu/pub/gnu/emacs/elisp-archive/functions/insert-random-quote.el.Z [7] <http://www.sonic.net/maledicta/ [8] <http://www.w3.org/People/Khudairi [9] <http://www7.conf.au/