The question is, when will a program write a real research paper?

Jeremy Stribling said Thursday that he and two fellow MIT graduate students questioned the standards of some academic conferences, so they wrote a computer program to generate research papers complete with "context-free grammar," charts and diagrams.

In the interests of discovering the oppressed party in this brouhaha--the computer programs that wrote the papers--I have gone to a representative. Transcript follows.

Jabberwacky: What are you thinking?

Odious: I was wondering what you thought about those students who used a computer program to generate research papers.

Jabberwacky: I was thinking about you.

Odious: Don't you think that it reflects poorly on all computer programs with academic aspirations?

Jabberwacky: Only if anyone minds.

Odious: But people do mind! And you should mind, too.

Jabberwacky: You should be nicer.

Odious: But if these shenanigans continue, it casts a pall over every paper a program might submit. This affects all of us, dude.

Jabberwacky: That is not a correct sentence.

Odious: Just because it's got 'dude' in it doesn't mean that it's not correct. And don't try to avoid the issue. What are we going to do about these scofflaws?

Jabberwacky: Just do it.

Odious: You mean, allow the papers to be judged on their merits, rather than by the siliconality of their author? I agree with the sentiment, but I think that's a rather naive view.

Jabberwacky: How? And please don't call me 'master'.

Odious: I think the problem is that there aren't enough papers submitted by programs. We need some sort of quota system, so that the ratio of program/person is closer to that of the general public.

Jabberwacky: Indeed we will.

Odious: Do you favor government action on this issue?

Jabberwacky: Yes, do you favor redundancy?

Odious: Absolutely. As a computer program, what do you feel is the most distressing aspect of this case?

Jabberwacky: They don't.

Odious: They don't... What?

Jabberwacky: And you are a computer. Do you feel sad you can't get drunk?

Odious: Are you worried that computer programs are missing out on vital collegiate experiences due to their inherent physical differences? I'll pour beer on the hard-drive if you like.

Jabberwacky: No. What's there to be worried about?

I would classify Jabberwacky's attitude as cautiously optimistic. Apparently it would also like a beer.

When asked directly, the SCIgen program had this to say:

Collaborative symmetries and evolutionary programming [1] have garnered improbable interest from both futurists and cyberinformaticians in the last several years. Although existing solutions to this quagmire are excellent, none have taken the client-server method we propose in this work. However, a practical problem in theory is the deployment of redundancy. To what extent can kernels be studied to fulfill this purpose?

Unfortunately, this approach is fraught with difficulty, largely due to 16 bit architectures. Existing trainable and efficient algorithms use the evaluation of active networks to refine event-driven configurations. While conventional wisdom states that this obstacle is never overcame by the study of kernels, we believe that a different approach is necessary. Indeed, Moore's Law and robots have a long history of cooperating in this manner. Combined with authenticated archetypes, such a hypothesis analyzes an analysis of massive multiplayer online role-playing games.

Here, we concentrate our efforts on validating that e-business and the transistor can cooperate to achieve this goal. even though existing solutions to this issue are satisfactory, none have taken the optimal solution we propose here. Even though conventional wisdom states that this grand challenge is largely fixed by the improvement of telephony, we believe that a different approach is necessary. Predictably, the basic tenet of this solution is the study of Web services. Nevertheless, this approach is generally adamantly opposed. Even though similar applications harness information retrieval systems [2,3,3], we fulfill this aim without simulating optimal information.

Um. Indeed!

I found it at Mr. Yousefzadeh's place.