3 ms·

My eyes were getting wider until package 4 — then I realized the joke. I was trying to figure out what DB autocite scanned…

parsimo2010 7y ago
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.
- grenoire 7y ago
  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).
  - trombonechamp 7y ago
    You should be able to do this with \write18.
  - mehrdadn 7y ago
    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.
- waynecochran 7y ago
  I think what made me more gullible is all the modern claims of machine learning... maybe, I thought, LaTeX was now using RNN's...
bArray 7y ago
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.
- jrochkind1 7y ago
  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.

parsimo2010 7y ago

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.

grenoire 7y ago
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).
- trombonechamp 7y ago
  You should be able to do this with \write18.
- mehrdadn 7y ago
  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.
waynecochran 7y ago
I think what made me more gullible is all the modern claims of machine learning... maybe, I thought, LaTeX was now using RNN's...

bArray 7y ago

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.

jrochkind1 7y ago
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.