Previously, NewScientist reported on a research project at Google, moving from their PageRank system to (for lack of a better word) TruthRank. Now NewScientist’s Feedback (21 March 2015) (paywall) column reports on a new hazard to be scaled:
RANKING internet search results according to how well they reflect “facts the web unanimously agrees on” (28 February, p 24)? What could possibly go wrong?
Adrian Ellis wrote to ask about something the web is unanimous on: that glass is really a very thick liquid (14 March, p 54). Well before the ink was dry on his letter, a colleague asked a FWSE (famous web search engine) about this.
It shot back: “Antique windowpanes are thicker at the bottom, because glass has flowed to the bottom over time… Glass is a supercooled liquid. Glass is a liquid that flows very slowly.”
At first glance, the web page it was quoting seems to exemplify a core Feedback hypothesis: that the internet holds many false beliefs, the more fruitloopy of which we catalogue.
However, closer inspection reveals a different flavour. The above text is from a page (bit.ly/GlassLegend) entitled “Glass: Liquid or Solid – Science vs. an Urban Legend”. It is in fact debunking the ideas quoted. This is a prime example of artificial stupidity. The search ranking system cannot tell the difference between myths and statements that mention them.
