Sunday, July 17, 2005, 12:16 PM
A few minutes after 5 pm, the scheduled starting time for our fake talk session, the conference room was quiet. And empty. Quiet and empty. WMSCI and the hotel had posted sentries in front of our room, redirecting anyone looking for WMSCI registration to their room across the hall. Well, we thought, no point in wasting a conference room and a video camera -- might as well get started, even without anyone in the audience.
Just as we were getting started, we got a first attendee. It was Allen, the man Max -- er, Dr. Thaddeus Westerson -- had spoken to earlier about the Institute for Human Understanding. Excellent. We called the session to order.
The format of the session was that each of the three of us would give one fake talk. None of us had seen anyone of the slides before; Jeremy had generated them that morning using SCIgen, and hadn't looked at them. We would all be in disguise, and the disguises would get more and more ridiculous as the session progressed.
The talks:
- Max Krohn as Dr. Thaddeus Westerson: Harnessing Byzantine Fault Tolerance Using Classical Theory
- Jeremy Stribling as Dr. Mark Zarqawi: Synthesizing Checksums and Lambda Calculus using Jog
- Dan Aguayo as Franz T. Shenkrishnan, PhD: On the Study of the Ethernet
The talk session was pretty amazing. Poor Allen stayed the whole time, and seemed utterly confused for the first two talks; by the end of Dan's talk though, he was laughing along with us. We do not know if he ever figured out they were fake, but he definitely knew that we weren't being entirely serious. Throughout the talk, people were coming in and out, mostly out. They grabbed a cookie or a soda, watched for a few minutes, and left probably still thinking they were watching a WMSCI session.
We had a great time giving and watching the talks. The highlight was definitely Dan's talk -- watching him try to keep a straiht face while adjusting his constantly crooked mustache was surely one of the greatest comedic moments of the 21st century. Check out the highlights from the talk in video form on our homepage.
In the end, we gave away a handful of cookies, 3 sodas, and a bottled water, and had greatly confused a small number of WMSCI attendees. We had accomplished our goal of giving randomly-generated talks at WMSCI, even if it wasn't in a real WMSCI session. As Dan proclaimed at the end of the session, "It hasn't been a total failure." And that's the best we could have hoped for.
--Jeremy
| permalink
Wednesday, July 13, 2005, 09:23 PM
Early in the morning of July 10th, 2005, we landed in Orlando, armed with a laptop, a projector, a USB key full of unseen randomly-generated talks, 150 fliers and a poster for our special Analogic Labs-sponsored technical session, and a backpack full of wigs and mustaches. Hurricane Dennis was threatening to slam the Gulf Coast of Florida and possibly points eastward, but we did not heed the doomsday warnings of weathermen. We were there with a purpose, one so great it dwarfed any concerns of personal safety. We flew toward a hurricane to give a fake talk in front of strangers at a spamference.
First things first: we needed a video camera. We rented a car and drove out to see Tadd Torborg, a loyal supporter who lives in the area. He and his welcoming family lent us a very nice digital camcorder and tripod, essential components of our operation.
After a quick lunch we headed to the Rosen Centre Hotel, located on
In the room: schoolroom seating for fifty people; complementary Rosen Centre pencils and brochures; a water pitcher and a bowl of candies on each table; a projection screen. Outside the room: a registration desk; a sign announcing the randomly-generated name of our session; an easel for our poster; and a perfect view of the WMSCI 2005 registration room, which was right across the hallway from our room. Entirely coincidental. Such luck cannot be taught.
WIth three hours to go before the official start of our session, we set up our poster and began handing out fliers to WMSCI attendees after they finished registering. Though most took the flier politely and then threw it away on their way to Disney World, some seemed genuinely interested in the technical content promised by the randomly-generated talk titles on the flier. "Byzantine Fault Tolerance" in particular seemed to catch the eyes of people. Max got caught up in a conversation with someone who assumed he was Dr. Thaddeus Westerson, and asked him about the Institute for Human Understanding. It seems this place exists in upstate New York, near Cornell. This person promised to come to the session and handed Max his amazing paper on political cleavages before heading off. It is impossible to convey how ridiculous, but bizarrely genuine, this paper is, but our colleague Russ came close: "It is indistinguishable from random content."
While we were handing out fliers and setting up equipment, we began catching dirty looks from the WMSCI folks across the hall. Nagib Callaos had a surprising number of lackeys and "PC members" on hand for registration; though probably it shouldn't be surprising that such a deceitful organization would have to be well-organized. We believe we caught a fleeting glimpse of Nagib Callaos in the room at one point -- short, bald, well-dressed -- but he's a wily one. As soon as an attendee brought one of the fliers to his attention, he fled to another conference room and closed the door, along with several members of his entourage. They began sending very obvious spies over to get fliers and talk to us, and posted sentries to keep an eye on us. We were officially on the radar.
A bit later a member of the hotel staff approached us. She was the one who had worked with Mr. Barry Swale of Analogic Labs to organize our conference. She was a very nice woman, and we felt bad that she was caught in the middle of this storm of controversy, but a bit of collateral damage is inevitable in any undertaking of this magnitude. She politely asked us if we were "soliciting attendees from other conferences." Well, we were actually inviting anyone that walked by to attend; anyone was welcome at the 6th Annual North American Symposium on Methodologies, Theory, and Information. "The problem is," she said, "it says WMSCI right here on your flier." Why did that matter? "When I asked Mr. Barry Swale if this conference was in conjunction with any other conference happening at the hotel, he said no."
Here is where things get a bit confusing. Apparently, in the language of hotels, "in conjunction with" means "in the same field as." This is not obvious. After a lengthy argument on semantics, she decided that we could stay at the hotel, provided we remove all mention of WMSCI from our fliers and poster. She took a flier and produced for us copies that entirely excluded the first paragraph. Eventually, she told us we could no longer hand out fliers, or even talk to WMSCI attendees. Somehow, we managed to be both in conjunction with and in competition with WMSCI all at the same time, and this was forbidden by hotel policy (even though our contract stated nothing of the sort). She even tried to move us to a different conference room down the hall, but we refused and were firm in our resolve.
With a few minutes to go before 5 o'clock, they brought in our refreshments. It seems that cookies for 25 people (at $9/person) actually translates to a total of 29 (somewhat large) cookies. Wow. Our refreshments, projector and video camera were in place; all we could do now was wait and hope that a few of the people we had talked to earlier would actually show up for our session. The hotel and WMSCI placed a few people outside of our conference room, directing WMSCI attendees into the registration room, so they would not become confused and wander into our room. Curses!
But, again, we were firm in our resolve, and were determined to give the talks, even if it was to an empty room. We waited there, costumed and pacing, in our empty room, the eye of the storm, ready for whatever was next.
--Jeremy
| permalink
Thursday, July 7, 2005, 01:52 PM
July 5, 2005: After a lot of work expanding the talk generator (adding clipart, graphs, extra slides, etc.), I decided it was time for another practice talk. With just a few days left before heading down to Florida, I gave a randomly-generated talk at the weekly meeting of my research group (PDOS) at MIT.
Of course, everyone in the group knew it was going to be a fake talk, so I wasn't trying to fool anyone this time. I simply needed some practice looking at slides for the first time and coming up with explanations for them on the fly. The whole thing went pretty well, although it was very difficult keeping a straight face while everyone in the room is laughing at what I was saying. Especially when explaining how we used "KGB's 2-node testbed" to disprove the work of "Swedish physicist Venugopolan Ramasubramanian".
The slides for the talk are here.
There is also an audio recording of the talk here [12 MB] (thanks to Thomer Gil!).
--Jeremy
| permalink
Thursday, July 7, 2005, 12:18 AM
So we had this problem. What we thought would be a little website that would be visited by friends, and maybe friends of friends, had basically been seen by everyone in the known universe within a few days. We had become more powerful than anyone had every been in the history of mankind. And with great power comes great responsibility, or something. People from all over the world donated money to us so that we could attend WMSCI and give a talk, and we desperately wanted to do so, but we had been unvited. So what were three simple grad students to do?
Our initial reaction to the news that we could not present Rooter at WMSCI: use someone else's slot at the conference to give the fake talk. Surely someone who had a paper accepted at WMSCI had seen our website, and realized now that the conference was sort of a scam. This hypothetical person should be more than happy to give up his/her talk, and let us give a fake one in its place. The plan was brilliant; even if we publicly asked for a volunteer to surrender their slot and Nagib Callaos heard about it, he would never know just which talk at WMSCI would be fake. He'd have to shut down the whole conference if he wanted to stop us.
So we put up a notice on the site, sat back and waited for the offers to roll in. And waited. One person emailed saying he would be willing, but then emailed back and said his advisor would not let him get involved in a "feud" between "researchers." Hmmm. And waited.
As we waited, we started considering things from a different angle. Would we be able to get a badge under someone else's name, or would we just sneak into this conference to give the fake talk? Was it legal to video tape such a talk at this conference, without permission, and post the video to the Internet as we had promised? Did WMSCI hold the copyright to everything that happens within their conference? Was it right morally and ethically?
Then Dan came up with a different idea, during a phone interview with a magazine reporter. What if we held our own conference, in the same hotel as WMSCI and at the same time? And we somehow disguised it to look like a real session of the extremely multi-tracked WMSCI? That way we could get real conference attendees there, video tape everything in the open without fear, and give as many fake talks as we wanted. The plan would originally publicized would remain simply as an elaborate cover. Of course there would be a ridiculous amount of planning and preparing for this plan, but what better way to spend the precious few months befoire your Masters thesis is due?
Of course we couldn't just book a room at a hotel under our real names -- Nagib Callaos might find out and ruin everything. But we couldn't use random fake names either, as there would surely be contracts and credit card transactions involved. Then our friend Sean suggested something that was perfect: registering a real company with the city as a DBA ("Doing Business As"). You can totally walk into a bank and get a credit card with such a thing. USA! USA! USA!
But we needed a good name. One full of inside-joke goodness, but which would not raise any red flags down at Callaos headquarters. We turned to a tool that man had been employing for centuries to conceal his subliminal messages: the anagram. To the Internet Anagram Server! We entered the name of our favorite spamference patriarch, and after sifting through the pages of results, found the perfect name for our new company: Analogic Labs. Within two hours we registered as a DBA with Cambridge, MA, and opened a bank account. We were in publishing. We had a domain name. We had a logo:
;){kind=logo}
Logo credit: Sanjit Biswas, Frank Dabek, and Jeremy Stribling.
Jeremy started making the necessary arrangements with the Rosen Centre in Orlando as Barry Swale, the event planner for Analogic Labs. We learned that planning a conference is damn expensive (i.e., $9/person for a plate of cookies => 1.5 cookies/person), but we had plenty of money thanks to our many Internet supporters. We signed contracts, booked plane tickets, made fliers. We registered professional-sounding gmail accounts while talking on the phone to someone wanting to email us hotel menu information. We kept our plans secret from almost everyone.
But answers tend to lead to more questions. Would anyone even show up at our conference? Would they realize, having heard about SCIgen, that the conference was fake? Just how far could we push this thing? Now, with the conference just three days away, we still don't know the answers. But one thing's for sure: hotel cookies are mind-blowingly expensive. God damn.
--Jeremy
| permalink
Tuesday, July 5, 2005, 10:04 PM
Once upon a time, an MIT professor named Mary Cummings contacted me regarding SCIgen. She said that she would donate to our cause, but on one condition: I had to spend the night in the haunted mansion on the hill.
Ok, that was a lie. The real condition was that she wanted to chat with me sometime about the SCIgen program. This was after donations were closed, but I said I'd be happy to talk with her anyway. After a few email exchanges, she invited me to give a talk at her cognitive complexity seminar in the Aero/Astro department at MIT. The students had discussed SCIgen in class when it first came out, and she thought it would be interesting for me to come in and talk about how the program works, and have a class discussion on the academic ramifications. I said I would do it, but on one condition: the class had to be held in the haunted mansion on the hill.
Another lie. My real condition was that I got to use the first five minutes of my visit to give a practice randomly-generated talk. This would be a great forcing function for me to get started on the talk software, and an opportunity to practice not cracking up when reading phrases like, "Our implementation of our approach is low-energy, Bayesian, and introspective" out loud. She agreed not to tell the class that I was coming beforehand, and I got to work on the software. I didn't have time to put together a complete talk with figures and graphs, but I was able to put together a talk generator that produced four-slide talks, laced with words pertaining to the subject of the class: cognitive complexity.
May 11, 2005: She introduced me to the class of 10 students as a "special guest speaker", and I gave them a psuedonym and said I was a research scientist at CSAIL "working in that gray area between computer science and cognitive complexity." Then I started up my talk, having never before seen the random slides being shown on the screen.
The talk? "The Relationship Between the World Wide Web and Unmanned Systems.". I told the students to interrupt anytime they had questions or something wasn't clear, and then I started off. Things looked bad when someone raised his hand on the very first bullet, asking what the "Yerkes-Dodson Law" was. Luckily, I had taken this from a cognitive complexity paper that I had skimmed the night before , and sort of remembered what the graph describing the law looked like.
"Did you not cover this in class?" I said. "It's the one that looks like this." And I waved my hands around. I looked to the professor (who was doing a great job of not laughing) for help.
"Yes, we did cover that," she said. "It's the upside-down curve." The class nodded in recognition. Crisis averted!
Needless to say, after that smackdown, there were no more questions through the remainder of the talk. I seemed to be keeping their attention for the most part, and no one fell asleep for those five minutes. When I got to the end of the fourth slide, I paused for questions. "Does anyone have any questions? Is there anything that's not clear?"
*Cricket*. No one spoke up. I waited a beat. "Well, I have a question for you," I said. "Did anyone actually think this was a real talk?" I watched realization dawn slowly over their faces.
"Wait," said one student. "Are you that guy who wrote that fake paper?"
I fessed up, and everyone breathed a sigh of relief. They said they just assumed the talk was real, and that I knew what I was talking about, since I was using buzzwords in their field. They had hoped that eventually they would begin to understand what I was saying, presumably by the middle of the talk when I started to get to the point. Then we had a great discussion about the program, and how bad it is for academics to just sit back and not ask questions in the face of confusing scientific exposition.
Afterwards we all spent the night in the haunted mansion on the hill, and there were ghosts and it was scary but we solved the case with some excellent team work. And we lived happily ever after.
--Jeremy
| permalink
Tuesday, July 5, 2005, 06:12 PM
At some point during all the press craziness, Science magazine asked us for a picture of us for a SCIgen story they wanted to do. We asked our good friend Frank Dabek to take a few of us. Here's one we took in front of the Department of Systemics, Cybernetics, and Informatics at MIT's CSAIL laboratory.
;)
Left to right: Dan Aguayo, Max Krohn, Jeremy Stribling. Photo credit: Frank Dabek. Sign credit: Sanjit Biswas.
Oh, and here's the picture they eventually used in Science:
;)
Left to right: Dan Aguayo, Jeremy Stribling, Max Krohn. Photo credit: Frank Dabek.
--Jeremy
| permalink
Tuesday, July 5, 2005, 06:02 PM
In late March 2005, we were informed by Nagib Callaos Himself that our Rooter paper had been accepted as a "non-reviewed paper" to WMSCI 2005. The best part about this acceptance email was the fact that it was caught by the gmail spam filter -- it wasn't until several days later that we discovered it. When even the acceptance notification to your spamference is considered spam, you know you're doing something wrong.
So the impossible had come to pass. Even though it was clear from the acceptance email that no one had read the paper (i.e., it hadn't actually fooled anyone), it looked enough like a paper that they had accepted it by default. This is more than can be said for the Mazières and Kohler paper, which to our knowledge was never officially accepted or rejected from WMSCI. Obviously, we were ecstatic; but now, what was the next step? Did we dare carry this project all the way to its logical extreme: presenting the paper with a randomly-generated talk at WMSCI?
The answer, clearly, was an emphatic yes.
But the conference was in Florida, and we were simple grad students way up in Boston: how could we afford to make the trip? Moreover, the conference wanted $390 to publish the paper and attend the conference. It would be pretty difficult to convince our usual source of conference travel money (our advisor's research grants) to pay our way. So we turned to the one thing we trusted to provide us with wisdom, guidance, and kindness: the Internet.
Using his HTML non-skillz, Jeremy set up a simple website with the WMSCI story, the SCIgen software, and a PayPal donation button. This might have been interesting enough to attract a little attention, but it was missing something . . . So, despite all of his numerous social obligations, Jeremy spent a day or two wrapping a web interface around the SCIgen software, allowing anyone to create their own papers with a click of a button. Then on the evening of April 11, 2005, fingers crossed, he sent five or six emails to friends and colleagues, hoping for a few donations over the coming months to help fray the costs of the WMSCI trip.
From these humble beginnings: Metafilter! Fark! Slashdot! BoingBoing! Boston Herald! Front page of CNN! Front page of BBC! $2400 in three days!
Incredible place, that Internet. Before we knew what hit us, our servers were melting, our inboxes were full, and our PayPal account was overflowing. The reaction was very supportive, and everyone loved the idea of heading down to Florida to give a talk. We're still blown away by all the attention.
Of course, fame is a double-edged sword, laced with trace amounts of irony. Reporters began trying to hound Nagib Callaos for quotes; Internet citizens started flame-emailing him; surely the WMSCI servers saw quite an unusual spike in traffic that week. When we finally had enough money to register for the conference, we did so; but by then it was too late. Nagib refunded our money, and told us we were no longer welcome at his spamference.
This saddened us deeply and profoundly. The sense of loss was palpable. Had we destroyed our chance at giving a fake talk by coming public too soon? Gerald Sussman, venerable MIT professor and noted prank connoisseur, publicly berated Jeremy in his office for exposing the "hack" prematurely. To paraphrase Sussman: "You needed money? You should have just asked me! I could have gotten together three professors to pay your way, no problem!" And he threw his hands up and stormed out. Damn, if only we had thought of asking Gerald Sussman for $1100 . . .
Well, anyway, the cat was out of the bag, and there was nothing we could do but soldier onward. We were more determined than ever to go to the conference and give a randomly-generated talk. But that was off in July, and there was still much work to be done.
--Jeremy
| permalink
Friday, July 1, 2005, 05:24 PM
Welcome to our unbelievably pretentious and self-serving collection of antecdotes about SCIgen. This is not really supposed to be, in the parlance of our times, a "blog"; it's just a place to write down information pertaining to a simple, paper-generating program.
SCIgen was born in February 2005. Jeremy and Max were sitting around in their office at MIT one Sunday afternoon, doing their best to avoid any real work. David Mazières and Eddie Kohler had just submitted their seminal work to WMSCI 2005, a conference known throughout the academic CS community as one that sends out large amounts of spam soliciting submissions. Headed by multi-conference magnate Nagib Callaos, WMSCI sends out numerous emails addressing recipients as "Dr.", and was notorious for refusing to remove you from their lists. Mazières and Kohler's submission was a brilliant tactical response. The gauntlet had been thrown.
Max, inspired by Chris Coyne's now-defunct term paper generator, proposed the idea of an automatic research paper generator. It would be the ideal weapon in the war against 'spamferences' like WMSCI, allowing the instantaneous creation of unlimited paper submissions. Plus, it could be really, really funny.
He set to work modifying the original term paper generator, eventually rewriting it from scratch in Perl. The scripts were surprisingly simple; it soon became clear that the real bottleneck would be writing a gigantic grammar, big enough to generate sufficiently-random and diverse papers at least a few pages long. So, Max and Jeremy toiled away for a few weeks on the grammar, drawing inspiration from a number of real papers, but writing everything by hand.
As the WMSCI 'submission deadline' (it turns out this was extended a number of times) drew closer, Jeremy became insane and started work on random graph and figure generators, giving up all pretense of doing any real research. However, it soon became clear that more help was needed, and as the final hours ticked by, Dan volunteered his not-inconsiderable talent for bullshitting.
After several iterations of generating trial submissions and subsequently fixing bugs and tweaking the grammar, we generated "Rooter: A Methodology for the Typical Unification of Access Points and Redundancy." It was perfect in every way. We also submitted a second paper, "The Influence of Probabilistic Methodologies on Networking," under the name of our labmate Thomer.
After submitting, we returned to our normal grad student lives. Parting with SCIgen was bittersweet, and everyone was melancholy, but we were stronger people for it. As the weeks went by, and the excitement surrounding the WMSCI submissions faded and was replaced by more pressing and legitimate research concerns, SCIgen sat gathering virtual dust in the humble closet of our CVS repository. We were but mortal men, toiling away in anonymity at our daily lives, inching ever closer to that mythical degree, unaware that soon, Rooter would return.
--Jeremy
| permalink
