The best part of the joke is that this is somewhat plausible. It would be relatively simple to write a package that threw in a “nocite” of your leadership’s key papers so they show up in the references even if you don’t cite them in your paper.
The translate package could be implemented with Google Translate or similar. The INTERCAL language required saying “please” in your code, something similar could be done for the boast package. I bet someone sufficiently motivated could make a decent stab at most of these, except for the award and accept packages. Some might require a few config options though.
Translate was the one that threw me off immediately, and then I caught on the joke. To my knowledge, you cannot make HTTP requests with LaTeX (maybe some processor has commands to do so, alas I don't know of any).
Nope, I actually found that one possible. There's already a package (minted) for syntax highlighting, and it calls external programs (the pygments Python script IIRC), so I don't see any reason why they couldn't do that.
Yeah, i was thinking "huh, I guess you could build a Markov model to figure out a string of keywords that leads to a given citation, then auto-insert for the author". All you would need to do is tell it to load a certain database/package for a given research area.
Given how many papers are online you could quite "easily" build up a nice little database. Add in some checks for quality and you've got yourself something that can make a reasonable guess for you. You would probably want some way to stop it from adding incorrect citations, but other than it would probably generally help.
As long as you citing things you haven't actually read is considered good practice (hint: Turns out it is!).
I think it's probably harder to do this tech than you think; on the other hand, I am continually amazed at the acceptance of throwing whatever citations you haven't read that you want in, so the bar may be low.
I actually wrote code for the autofit. Didn't do multiple pages, but it would adjust the size of a text to fit it into a given space (it was for printing chemical bottle labels). That was a fun project. I had a customized font set made by Bitstream for setting the Japanese text (this was years before Unicode was a viable thing, so we were encoding non-Latin scripts using SJIS) and had to go through a couple hundred pages of character tables to verify that we didn't have any coding errors when we switched to that font from the free bitmapped font we had previously.
Reading the first package: I really need this.
Reading the second package: Wow this is amazing, but slightly implausible.
Reading the third package: Ah fuck.
22 comments
[ 5.3 ms ] story [ 59.0 ms ] threadThe translate package could be implemented with Google Translate or similar. The INTERCAL language required saying “please” in your code, something similar could be done for the boast package. I bet someone sufficiently motivated could make a decent stab at most of these, except for the award and accept packages. Some might require a few config options though.
Given how many papers are online you could quite "easily" build up a nice little database. Add in some checks for quality and you've got yourself something that can make a reasonable guess for you. You would probably want some way to stop it from adding incorrect citations, but other than it would probably generally help.
I think it's probably harder to do this tech than you think; on the other hand, I am continually amazed at the acceptance of throwing whatever citations you haven't read that you want in, so the bar may be low.
https://web.archive.org/web/20080705135647/http://www.jmlg.o...
A small number of old NIPS (now NeurIPS) papers may even have used it.