Expert goes bust

carnivas
Little world of carnivas
2 min readNov 28, 2017

--

Often, I see a pattern play out, in this order:

  • There is a person, whose writings I read or whose speech I hear (or some such thing), get mighty impressed with the insights they have to offer
  • Slowly, it goes to a stage that I form a great positive impression of this person and go recommend to everyone to read those writings — ‘You can read anything this person writes’ I say
  • Time passes
  • Then comes a time when this person produces a piece somewhere on a topic that I understand well enough
  • And I find that piece too naive or outright wrong, but presented in a very confident manner
  • First, confirmation bias kicks in and I try to justify my case of this person being great by saying that this was a one-off piece where things did not work out
  • Slowly, I realize that there is no way for me to know how correct this person would have been in the other topics, which was anyways new to me.
  • I feel a bit foolish (gullible) to have trusted just the confident words

I am not sure if there is a word for this ‘feeling’.

With whatever self-awareness I can muster, I am confident this is not the other case of ‘confirmation bias’, where I agreed with the person when the writings conformed to my views and found it naive when it went against my views. Sometimes, even plain facts are presented wrong but in a confident manner.

This has happened across domains — Indian political analysts (till they talk about Tamil Nadu), tech industry commentators (till they talk about India), old-economy strategy experts (till they talk about latest developments in tech) and so on.

So, one of the mental models to address this is to have a filter applied whenever I get this feeling of ‘Read anything this person writes’, temper the enthusiasm a bit and fine tune the BS detectors.

Update:
This is a variant of the “Gell-Mann amnesia” effect (feel free to call it the Carnivas Amnesia Effect) that I refer to in this post on mental model for news consumption:

--

--