Now put it in an algorithm that detects any mention of "only, always, never" or any other absolute. Then insert a boilerplate at the top of such discussions, saying there are always exceptions, but the gist of the argument is still valid ;)
... not that I am without guilt on this ...
So now I actually appreciate more confident, strong language, because when I read a blog post or article like this, I can take the caveats as being there, but unsaid.
I sometimes think that if everyone took this assumption, Hacker News would decimate its comment count!
As a side-effect of his work on expert forecasting[1], Tetlock discovered that people put more stock in those who made confident predictions than those who qualified their predictions. To the point that he basically says: if a person is a famous pundit, they are more likely to be wrong, because the media prefers bold predictions to maybes and perhapses.
In business you can be bold or not bold. If you're not bold, you are at a social disadvantage.
Then you can be right or wrong. If you are right, everyone hails you as a visionary. If you are wrong, nobody remembers.
So in fact, in business, the rational strategy for an individual is to speak confidently and let the dice fly high.
[1] I reviewed his book a while back. http://chester.id.au/2012/07/29/review-expert-political-judg...
For example, I can't truthfully say that my code is bug free or the best or whathaveyou. But I can say "My product improves <business problem>."
I guess the trick is to have two hats. The first hat is the problem domain hat. This is where you make confident, accurate claims about your product.
The second is the solution domain hat. This is where you park your professional anxieties about whether you've crossed and dotted everything.
Qualifiers are substituted by getting multiple opinions from different pundits.
Luckily there's an extreme predictor for every extreme event, so simple selection happily promotes some loudmouths to guru every now and then.
This gives you the handles you need to grasp and wrestle with the vagueness of the actual world without either needing to give up rigour or rely on hacks and undocumented conventions about the mapping from the real world into the logical world.
If you've got the patience, I can give as an example a long review I wrote of Sidney Dekker's Drift into Failure, where in part I argued that his argument largely vanishes if reframed in terms of fuzzy logic: http://chester.id.au/2012/04/09/review-drift-into-failure/
Edit: Hmm, I really look like a spammer in this thread. Not intentional. No more than usual, anyhow.
I'm not trying to be a stickler. Granted, I work mostly in science research, where broad claims are discouraged and quantifying confidence in a hypothesis is expected. Extraordinary claims require extraordinary evidence.
At best, I think this rule could apply to (maybe) 80% of "huge" tech companies. 80% is a disingenuously long distance from "all companies".