> their published probabilities are predictions for what would happen if the event occurred today.
This is manifestly untrue. For the 2016 election 538 had a "Now-cast" that they very clearly marked as "Who would win the election if it were held today" (see here: https://projects.fivethirtyeight.com/2016-election-forecast/) and their other models incorporated the fact that there's time left before the election. Due to people not finding the now-cast valuable they abandoned it for 2018.
Nate Silver definitely incorporates uncertainty in the forecast (and talks about it frequently), but it's not modeled as a random walk. Rather they use historical data to look at what polling now has to say about an election however far away the election is. Taleb's criticism that they don't consider a black swan event might be valid, but Silver isn't just predicting the election now.
"People are jealous" seems to me like the uncharitable formulation. I'd phrase it more along the lines of: "we evolved in small tribes and are more acutely aware of relative status than absolute welfare." Most people's assessment of how well off they are comes from comparison with their neighbors rather than absolute knowledge of weather they have enough.
I agree though that eliminating severe poverty is a more worthy humanitarian goal than trying to produce equity.
In their model the only benefit of talent is it makes you more likely to double your wealth given luck. This means that they assume luck is needed for advancement while talent is not. Hence they assume that luck is more important than talent, and it's unsurprising that they see the result.
Likewise the power distribution is built in. Power distributions are created by multiplicative laws and the only ways wealth changes in this model are multiplication and division. This means they also assume the power distribution they claim to discover.
It's definitely very hard to build models that match reality, but this one doesn't really seem to try and it's outcomes are so predictable they don't offer meaningful insight. Luck surely plays a huge role in success, but something like longitudinal studies where we measure the abilities of children and then follow their outcomes are likely far more insightful.
Their model is if you have a lucky event you have P(talent) chance of doubling your capital and if you hit an unlucky event your capital is halved. Any step in time where you neither experience luck or unluck you are left unchanged. It seems wrong on face for several reasons:
- You need to be "lucky" in order to ever increase your wealth, which makes the "finding" that luck is more important than talent completely obvious from the way the model was designed
- Capital only changes by multiplication and division, so the fact that this model "recreated" the power distribution we see in wealth in the world is obvious from the setup. It was designed to be a power distribution and couldn't be anything else
- Talent has no effect on your life during unlucky or neutral periods in their model. This obviously fails to conform to the real world.
tldr; they designed a power-distributed model where luck was clearly more important than talent, and "discovered" that it showed a power distribution where luck is more important than talent.
Can't speak for many professions, but a long series of interviews where you do a bunch of modules and meet a large proportion of your team is pretty typical in consulting.
This is missing the last point of my argument though. Paid professionals like those that build Photoshop or Google are filtered by the hiring process.
Steemit (and I assume Woyano in the parent) is lacking both the GIMP/Wikipedia filter of people who are willing to work for free and the Photoshop/Google filter of screening people for ability before paying them to do something. This can land it in an unhappy valley where people are encouraged to churn out whatever and sheer volume makes it harder for good content to rise to the top.
I’m not saying it’s impossible for a model like this to work, just that it’s easy to fall into a trap where the best strategy to make money is to churn out low-quality content and so that’s what you get flooded with. Think of it as similar to the clickbait problem in online journalism.
Maybe the distinction lies in the filtering. Stackoverflow and Wikipedia have high quality because only people that really care about a subject (and thus usually know a lot) will invest time for free.
Once you start paying people, the much larger audience of 'people that want money' starts posting and the good people get drowned out.
Your friends who are employed were filtered by the hiring process. I imagine Google would be much worse if they hired everyone who applied.
I think the key distinction here is between “intolerance” and refusal of services. No one should be going out of their way to hear or publish the Nazi point of view, and people ought to call them disgusting and monstrous when they appear in the news.
That said we shouldn’t deny them services like hosting, because maintaing a neutral hosting stance is what allows for a firm argument that protects imprortant discourse like legitimate political dissent. Once you allow for a line to be drawn, suddenly bickering over that line becomes a an endless problem that might stifle any idea.
This is why I said "seem." Everything has some bias, but I think foreign publications usually have a less-relevant-to-me bias, which is about as close to unbiased as I can get.
I find the BBC and the Economist good for this reason. They are/seem a lot less biased than U.S. stuff through some combination of higher standards and that they'd rather manipulate the emotions of Europeans than Americans (and I am an American).
I feel like disappointing research is going to be a given when you have such an obvious conflict. The author has a strong motivation to make Zestimates seem bad (the less reliable the resources you have to buy/sell a home on your own are the more you'll need to contact them) and a million degrees of freedom (which neighborhoods in which cities, price floor for consideration, how many examples to pull ...) to get the answer they want.
"Research" of this type should probably never be trusted.
Fake news is a real problem, but this seems like an unfair no-win for Facebook from major media outlets. When they had more human editors on trending topics it was "Facebook is injecting their own bias into news and manipulating the public."
Now that it's more purely algorithmic it's "Facebook isn't policing content enough and making it too easy for fake outlets to manipulate the public."
I'm certain that when they follow up by cranking up machine learning to censor fake content it will be "algorithms don't stop everything fake and sometimes block real things, thus manipulating the public."
I'm very glad that where she wrote "Multiply that 0.6 percent chance of getting any given job by the 10 or so appropriate positions in the entire world, and you have about that same 6 percent chance of “success.”" she was in the regime where p*n ~= 1-(1-p)^n and her conclusion was still correct.
This is manifestly untrue. For the 2016 election 538 had a "Now-cast" that they very clearly marked as "Who would win the election if it were held today" (see here: https://projects.fivethirtyeight.com/2016-election-forecast/) and their other models incorporated the fact that there's time left before the election. Due to people not finding the now-cast valuable they abandoned it for 2018.
Nate Silver definitely incorporates uncertainty in the forecast (and talks about it frequently), but it's not modeled as a random walk. Rather they use historical data to look at what polling now has to say about an election however far away the election is. Taleb's criticism that they don't consider a black swan event might be valid, but Silver isn't just predicting the election now.