This quote is so very true, and it drives me nuts:
> “Society likes black and white,” he answers. But answers in science are not always so cleanly resolved. “We have to be careful because if we give the impression that science never says yes or no, says always maybe, then people say, ‘Well, then I should not trust science.’ It is very delicate how to give this message.”
It reminded me of some of the polls before the 2016 US Presidential Election. Nate Silver at the time gave Trump about a 30% chance of winning, and when Trump won, there were lots of "Even Nate Silver got it wrong!" comments. Most people really don't understand the difference between anything less than 50/50 and 0, and it becomes a real challenge to explain things when reality works in probabilities at a fundamental level.
I'm not sure that's the best example. Probability of precipitation (PoP) is simply a misleading measure by itself. It has a threshold of >0.01 in for what is considered "rain". So if there's only a 20% chance that you'll get a minimum 0.01 in of rain, you'd naturally expect a low chance for it to be be much higher, and hence would expect not to get wet that day. The problem here is that a guaranteed 10 inches of rain over 20% of the area would still give you a 20% "chance of rain", which would still result in a 20% PoP, which is not what you would want for a useful measure of rain.
Yeah, and then those same people conflate predictions with polls: "urgh are these polls that gave Clinton a 95% chance of winning? I'll just ignore it then".
As pointed out in the replies to Nassim's tweet, he chose a more volatile Silver model from 538 (NowCast) to suit his argument as opposed to Polls+Plus. While his thesis may be correct he chose to represent it in a biased fashion causing me to question his entire motivation in attacking Nate Silver.
It doesn't seem like "I'm trying to raise awareness about this" but more "I could do that too, but it's wrong, so you should ignore him." Given Nate's success during the previous election cycles I can't help but suspect that Nassim's ego is at play here.
Nate Silver himself admited that his early Trump "probabilities" were just gut numbers with nothing scientific behind them:
"Unlike virtually every other forecast we publish at FiveThirtyEight — including the primary and caucus projections I just mentioned — our early estimates of Trump’s chances weren’t based on a statistical model. Instead, they were what we “subjective odds” — which is to say, educated guesses. In other words, we were basically acting like pundits, but attaching numbers to our estimates."
"When Trump came around, I’d turn out to be the overconfident expert, making pretty much exactly the mistakes I’d accused my critics of four years earlier."
"There’s a danger in hindsight bias, and in overcorrecting after an unexpected event such as Trump’s nomination."
Unexpected event. Hmm, I think someone wrote a book on that, and called them black swans...
> "our early estimates of Trump’s chances weren’t based on a statistical model."
The Polls+Plus model that Taleb criticized wasn't an early estimate and it seems kind of strange that you'd conflate the two. Trump receiving the nomination can be argued as a 'black swan' event (although whether it was is debatable because it was observed as a possible, if unlikely, outcome before it happened), however Trump winning the election was not.
Your original post linked to a straw man attack by Taleb against Nate Silver based on paper Taleb wrote that is explicitly constrained to binary systems. When I pointed out this problem with Taleb's argument you created your own straw man citing Nate's words about a related, but different, statistical problem involving more than two choices. That is not a result that Taleb claimed and seems to be wholly outside of the scope of our original discussion.
Ultimately I think this is an ego thing for Taleb, he makes a habit of critically attacking [0] people who have fame in related fields without actually contributing anything to the discussion. Again, it's hard to see this as anything but Taleb's ego at work. By calling out the famous 'smart' people as wrong using arguments that are impregnable to people without highly advanced mathematic training he effectively increases his media clout without actually contributing anything to society.
Serous question: Has Nassim Taleb ever admitted he was wrong?
Edit:
I just read about how poorly Nassim Taleb did with Empirica Capital, [1] it seems very surprising that Nassim Taleb has such confidence undermining the work of others given that, objectively, his work during that period was a failure.
As Taleb would say, debate the best possible version of your opponents argument, not the one most convenient to attack:
"Take for instance the great Karl Popper: he always started with an unerring representation of the opponents positions, often exhaustive, as if he were marketing them as his own ideas, before proceededing to systematically destroy them"
What I was saying was in the spirit of GP, which mentioned that he is sick of hearing people attack "science", and mentioned Nate Silver.
I was pointing out that Nate Silver himself admitted of not doing "science" with respect with Trump, at least in part.
Sure, Taleb has an ego, just like Wolfram. Taleb can also be a jerk, just like Linus. Everyone knows that. Dismissing him because of that is just an easy way out.
Ereditato was worried about giving an interview, but I'd say the author (Ransom Stephens) did a good job of conveying the truth of the story. I feel nothing but sympathy for Ereditato. Imagine the pressure he faced, being in charge of this experiment with results consistently showing faster-than-light neutrinos, having to be the spokesman for such an enormous and expensive project with such crazy results. The potential for embarrassment due to the no confidence vote might have been exploited by a less decent writer, but Eriditato seemed less concerned with his own reputation than the proper telling of how science works. Many scientists hoped that something weird had actually been discovered, some new physics, so it was a big letdown for them when it was all traced to the timing mechanism, and those egos are probably still smarting from hoping for the still-impossible. There must be a lot of people who have an axe to grind, even now. It's good to be able to see into the human side of science, to see how things can go wrong and how it can affect people, especially those who are in charge of organizations. It must be great to be in charge when everything goes smoothly, but otherwise, uh oh.
I still find the effort at self-examination this team underwent to be heroic, especially considering the public pressure. This, to me, represents the very best science - work that goes to extreme lengths to find their source of error.
What they discovered after years of testing seems like the most obvious: a few nanosecond delay was found in the timing devices. I understand that it was difficult to time the timers, but that does seem like the first place to start.
from the text it seems like a cable was different between experiment and the validation process. Thing is, they did a further (bunched) experiment two months later, and there it was wrong again? I have a bit trouble believing that chain of events. Seems more likely to me that someone wasn't careful when checking it in between these experiments.
Whenever people say that there's new research that threatens to crumble the foundations of mathematics (like Banach–Tarski) or physics, I try to remind myself that this is mostly just academic. Don't get me wrong, it's important work. But bridges and buildings won't collapse as a consequence of accepting or denying the axiom of choice.
Time travel could collapse (or build) a great number of bridges and buildings.
Nuclear physics, a comparatively trivial academic accomplishment, did collapse a great number of bridges and buildings, and stood a very real chance of collapsing all of the bridges and buildings. It still might.
It will, it's now simply a numbers game. Let's not kid ourselves, one day a two nuclear powers will escalate and then we'll get to see if our doomsday predictions were correct ;)
Neutrinos are generally expected to travel a tiny bit slower than the speed of light. Nobody really expects them to travel at the speed of light even if we are currently unable to measure the tiny difference.
I was a physics student in 1987, when neutrinos were captured from a far-away supernova. I rushed into my professor's office and asked breathlessly: "Does the timing of the observations give us a limit on the mass of the neutrino?"
He calmly replied: Sorry, kid. Somebody already submitted a paper on that.
That raises an interesting thought: Does interstellar space have a non-unity refractive index? If there's any kind of matter out there, then I suppose the answer has to be yes. But I wonder if it's quantifiable.
The term 'speed of light' is somewhat confusing. Just to clarify: there's the universal maximum speed, c; this is the speed with which light travels in the vacuum; in a medium light happens to slow down, while neutrinos do not, the result of which is their moving not slower, but even "faster than light."
the speed of light is based on two specific values of the medium in which the photons are travelling. The permeability and the permittivity, one affects the magnetic the other affects the electric.
Hence by the change of these two values, the speed of light (magnitude) changes. Hence, if we are able to create media in which these values are appropriately specified, we can create media in which the specific speed of light is less than c0 or greater than c0.
It, therefore, may be possible to create a medium in which light will pass through faster than a vacuum.
> “Society likes black and white,” he answers. But answers in science are not always so cleanly resolved. “We have to be careful because if we give the impression that science never says yes or no, says always maybe, then people say, ‘Well, then I should not trust science.’ It is very delicate how to give this message.”
It reminded me of some of the polls before the 2016 US Presidential Election. Nate Silver at the time gave Trump about a 30% chance of winning, and when Trump won, there were lots of "Even Nate Silver got it wrong!" comments. Most people really don't understand the difference between anything less than 50/50 and 0, and it becomes a real challenge to explain things when reality works in probabilities at a fundamental level.