538 is No Longer a Credible Source

Maybe it would help if you substituted “right” with “worth paying attention to” – and, that you compare a modeling team to others, rather than just looking at any one approach in a vacuum.

538/Silver’s 2016 model asserted that the probability of Trump winning was much higher than what other major models were asserting. Trump went on to win that year. Therefore (especially after considering other, similar examples, like the 2022 midterms), as a first pass at least, the two “children” of that modeling team – Silver and 538 – are worth paying attention to.

I had a really amazing D&D character awhile back, a Hill Giant/Air Genasi cross. My PC was up against a Dragon/Yuan Ti cross. It was time for the final attack. I knew that if I rolled a 9 or higher on my attack, I’d win the fight. I said, “I’m probably gonna win,” and then I rolled a 3, and my Hill/Airy lost to the Duan Ti.

I wasn’t wrong. That’s just how odds work.

Why me? It was Thing.Fish who claimed the Silves predictions were “right”. Then he denied that they could be wrong, as he only gave a %. I was pointing out you can have it one way or the other.

Explain that to Thing.Fish.

Apropos of not much: 538 is eating Nate Silver’s dust right now. Seems really weird that Silver has put daily percentages on the general election since last week, while 538 is still silent since Biden’s step-down.

The bit I cited had taken the position that 538, representing ABC, was being conservative and less willing to commit. This is consistent with that. Wanting more data before running the model.

They are not “silent”- they are still posting polls, just no predictions.

and the most recent poll has Harris +4

Just to be precise: forecast not prediction. The point made respectively here is the difference between the two.

I’ve found that these threads really point out who is and isn’t a gambler, any sort of tabletop gamer (RPGs, war games, whatever), XCOM player, sports fan, or similar. All of the above know that low probability events can and do happen with regularity. Especially when you only get one trial for the effect to resolve, trying to make some larger point on that one event is folly.

That does seem to be the wall that posters are smashing into headfirst. Repeatedly. Without good results.

Their official stance is that Harris wasn’t officially the nominee. I think they were just buying time.

It’s fairly common. I recall seeing people struggle to understand the difference between “fortune telling” and “modeling” in climate change discussions. You could point them to pages in research where they modeled three different scenarios - here’s the case when CO2 increases at current rates, here’s if it lowers, here’s if there are a lot of volcanic eruptions, etc. - to try and help them understand that no one is trying to predict the future, that’s not what we’re looking at.

But we are trying to predict the future.

Just drawn as a well-informed sketch of the probable generalities, not e.g. precise rainfall distributions by county by week in the year 2060 or 2160.

Lotta people just cannot (or more often will not) accept the reality of general thinking.

Rule-based thinkers work bottom-up from fixed examples to rules. Science works top-down, from ideas to rules. Huge difference.

For the record, the 538 website no longer exists. ABC just shut it down. All traffic now redirects to the ABC News site.

Jesus, 538 and Monmouth Polling shuttered in one week.

I heard about 538 and Monmouth, but what happened to Jesus??

Cancelled. Too Woke.

He quit after one pole, I think.

Two, actually. In opposition to each other. But, they nailed it.

That gets more at the crux of it, I think

Geez guys, don’t leave me hanging here. Get to the point.