[PLUG-TALK] Finding useful papers

Keith Lofstrom keithl at kl-ic.com
Wed Mar 18 20:20:19 UTC 2020


On Wed, Mar 18, 2020 at 11:04:18AM -0700, John Jason Jordan wrote:
> Well, if anyone know the answer to that question they haven't published
> it yet. Or at least DuckDuckGo can't find their paper.

DuckDuckGo doesn't track you ... and doesn't much care
what you enter into the search box.  Like most free-as-
in-beer "search" engines, it has a few hundred thousand
pre-generated responses, and matches your query to one
of those responses.  Advertisers pay to be included in
those responses, which funds the search engine company.
They include just enough useful information to induce
most of their customers to come back.

Google Scholar, Microsoft Academic, and in this case
pubmed ( https://www.ncbi.nlm.nih.gov/pubmed/ ) are far
better search engines for academic questions.  Other
specialized academic search engines serve other fields.

Journal papers aren't mushed to pablum, thus more difficult
to digest.  However, even if you don't understand 80% of a
paper, the 20% you do understand is more useful than the
regurgitated swill that journalists and bloggers pump out.  

If I understand much more than 20% of a paper, I will 
learn more faster by choosing more difficult papers.  
When I find one helpful paper, checking the forward and
backward citiations with Google Scholar, or the corpus 
of a particular author with Microsoft Academic, connects
me to many more papers.

Typically the BEST paper is three to five steps away from
the first paper I stumble across.  And that's the paper I
spend hours studying to understand completely.

Keith

-- 
Keith Lofstrom          keithl at keithl.com



More information about the PLUG-talk mailing list