SciHub...

[nickschurch]

Come on, we know you want to! ;)

Similar to some of the others, it'd require reverse engineering the text from PDF...

[blahah]

Some practical considerations:

• in some jurisdictions this might be unlawful
• even where it isn't this might open users up to the possibility of legal pressure from litigious publishers
• the number of papers is too large to fit in a single datasource using our current technology (though in future versions the scale of 10s of millions of entries will be possible and fast)
• the diversity of PDF sources makes it particularly hard to reverse engineer XML coherently

And as per #6, I note that we can't officially create or sanction something which might open us up to legal bullying.

[blahah]

I will however leave this open for discussion - it is a unique resource and deserves to be discussed.