3 ms·

> It's just another opportunity for HN "experts" to pontificate about [X] using mostly incorrect armchair analysis. Isn't this just classic Gell-Mann amnesia?

> It's just another opportunity for HN "experts" to pontificate about [X] using mostly incorrect armchair analysis. Isn't this just classic Gell-Mann amnesia? All HN threads are like this. (Probably all threads everywhere on the internet are like this.) The next time you read a thread on [to skim the current front page] Facebook data, consensus protocols, Google hardware, facial recognition, death metal, or the British coastline...before reading the comments remind yourself "this is just another opportunity for HN 'experts' to pontificate about [X] using mostly incorrect armchair analysis"

aaron695 8y ago
> The Gell-Mann amnesia effect describes the phenomenon of an expert believing news articles on topics outside of their field of expertise As long as you question the news articles as incorrect and try and work out why it's incorrect, then it's not Gell-Mann amnesia. Which the parent has done in this case.
mherdeg 8y ago
> Isn't this just classic Gell-Mann amnesia? My favorite-thing about the Gell-Mann amnesia effect is that the coiner of the term himself, in the very speech where he coined the term, fell victim to that effect. To Michael Crichton's great credit, in the 2002 talk where he defines the term, he gives us an extremely specific, testable hypothesis we can use to evaluate his idea. He says that the "effect" is as follows: it is not worth your time to read the newspaper and all newspaper stories which contain any kind of "speculation" or "predicting" have no predictive value. The example he chooses is the 2002 United States steel tariffs. Most mainstream newspapers he reads have quoted experts whose consensus is that the then-contemporary 2002 tariffs will negatively affect GDP. His hypothesis is as follows: we will find in retrospect that these predictions were useless and US GDP will not clearly be damaged by the 2002 tariffs. Was he right? Well... Years later, there is consensus among experts that US GDP was in fact negatively affected by 2002 steel tariffs (links to sources at https://en.wikipedia.org/wiki/2002_United_States_steel_tariff https://en.wikipedia.org/wiki/2002_United_States_steel_tarif... ). Driven to attack the idea of trusting any experts who predict the future, Crichton's key example was readily disproven. I have often re-read Crichton's talk and have found myself wanting to project extra layers of meaning on top of his ideas; I find myself arguing on his behalf to try to change what he is saying to make it "more true". The words he uses are carefully chosen and seem like they add up to an interesting idea. It is true that people seem to believe information without factual basis when it's presented in the "news" format; it's also true that people seem to believe opinions they already agree with when presented to them in any format; and it is certainly true that people asked to make predictions with incomplete information sometimes guess wrong, yet their "best guess" predictions are sometimes presented without robust explanations of the uncertainty and what it means. All that is true. But it doesn't change the fact that the "Gell-Mann Amnesia Effect" was introduced by saying that the untrustworthy lamestream media needs to stop yammering already about those steel tariffs hurting GDP, which there is no way they could predict. (It appears the media was right on that one.) Worse than relying on anecdata, though, his talk does not propose a robust way of actually measuring whether media predictions are accurate -- he doesn't even try to measure that, let alone present results. (I have giggled at this before - https://news.ycombinator.com/item?id=18005236 https://news.ycombinator.com/item?id=18005236 )
- perl4ever 8y ago
  "He says that the "effect" is as follows: it is not worth your time to read the newspaper and all newspaper stories which contain any kind of "speculation" or "predicting" have no predictive value." To me, "no predictive value" is consistent with sometimes being right, possibly by chance. It means being wrong enough that you can't assume either positive or negative correlation with the truth. It's easy to hide statements which are true and should be obvious in a sea of confusing nonsense.
rootusrootus 8y ago
Yes, absolutely! I actually had the same thought yesterday and I walked away from HN for a break, grumbling to myself about how true the old joke is about computer programmers having so much hubris that they think they are smart enough to solve every problem in the world ;-). See also Soylent. Yesterday it was just an overload because we must have had a half dozen 737MAX threads in the previous 24 hours. And the quality of the commentary was going down, not up, as time went on. Totally agree with your advice to keep in mind that this applies to all topics. A little part of me keeps whispering that I should avoid looking at the comments altogether. I rationalize that on HN I am getting good quality commentary and the first few contrary arguments will help me see the article/news/whatever in an objective light, but then I totally get sucked in.

aaron695 8y ago

> The Gell-Mann amnesia effect describes the phenomenon of an expert believing news articles on topics outside of their field of expertise As long as you question the news articles as incorrect and try and work out why it's incorrect, then it's not Gell-Mann amnesia. Which the parent has done in this case.

mherdeg 8y ago

> Isn't this just classic Gell-Mann amnesia? My favorite-thing about the Gell-Mann amnesia effect is that the coiner of the term himself, in the very speech where he coined the term, fell victim to that effect. To Michael Crichton's great credit, in the 2002 talk where he defines the term, he gives us an extremely specific, testable hypothesis we can use to evaluate his idea. He says that the "effect" is as follows: it is not worth your time to read the newspaper and all newspaper stories which contain any kind of "speculation" or "predicting" have no predictive value. The example he chooses is the 2002 United States steel tariffs. Most mainstream newspapers he reads have quoted experts whose consensus is that the then-contemporary 2002 tariffs will negatively affect GDP. His hypothesis is as follows: we will find in retrospect that these predictions were useless and US GDP will not clearly be damaged by the 2002 tariffs. Was he right? Well... Years later, there is consensus among experts that US GDP was in fact negatively affected by 2002 steel tariffs (links to sources at https://en.wikipedia.org/wiki/2002_United_States_steel_tariff https://en.wikipedia.org/wiki/2002_United_States_steel_tarif... ). Driven to attack the idea of trusting any experts who predict the future, Crichton's key example was readily disproven. I have often re-read Crichton's talk and have found myself wanting to project extra layers of meaning on top of his ideas; I find myself arguing on his behalf to try to change what he is saying to make it "more true". The words he uses are carefully chosen and seem like they add up to an interesting idea. It is true that people seem to believe information without factual basis when it's presented in the "news" format; it's also true that people seem to believe opinions they already agree with when presented to them in any format; and it is certainly true that people asked to make predictions with incomplete information sometimes guess wrong, yet their "best guess" predictions are sometimes presented without robust explanations of the uncertainty and what it means. All that is true. But it doesn't change the fact that the "Gell-Mann Amnesia Effect" was introduced by saying that the untrustworthy lamestream media needs to stop yammering already about those steel tariffs hurting GDP, which there is no way they could predict. (It appears the media was right on that one.) Worse than relying on anecdata, though, his talk does not propose a robust way of actually measuring whether media predictions are accurate -- he doesn't even try to measure that, let alone present results. (I have giggled at this before - https://news.ycombinator.com/item?id=18005236 https://news.ycombinator.com/item?id=18005236 )

perl4ever 8y ago
"He says that the "effect" is as follows: it is not worth your time to read the newspaper and all newspaper stories which contain any kind of "speculation" or "predicting" have no predictive value." To me, "no predictive value" is consistent with sometimes being right, possibly by chance. It means being wrong enough that you can't assume either positive or negative correlation with the truth. It's easy to hide statements which are true and should be obvious in a sea of confusing nonsense.

rootusrootus 8y ago

Yes, absolutely! I actually had the same thought yesterday and I walked away from HN for a break, grumbling to myself about how true the old joke is about computer programmers having so much hubris that they think they are smart enough to solve every problem in the world ;-). See also Soylent. Yesterday it was just an overload because we must have had a half dozen 737MAX threads in the previous 24 hours. And the quality of the commentary was going down, not up, as time went on. Totally agree with your advice to keep in mind that this applies to all topics. A little part of me keeps whispering that I should avoid looking at the comments altogether. I rationalize that on HN I am getting good quality commentary and the first few contrary arguments will help me see the article/news/whatever in an objective light, but then I totally get sucked in.