re: Nate Silver has lost it and never be taken seriously again

quote:

---

Do you think that a model which results in Biden getting 413 EVs as the most likely result is the same thing as a prediction that he will get 413 EVs?

---

Yes, otherwise there is no real purpose to doing a model of anything.

quote:

---

Or, you know, inform.

---

Of course they provide information.

It's kind of hard to persuade and/or manipulate without saying something.

When I use models in my business I am always trying to bring about a group consensus decision. Models are entirely adjustable in many directions.

quote:

---

Or, you know, inform.

---

"2 million dead from COVID"

"Seas will rise by 2 feet by 2020"

Muhhhh models

quote:

---

quote:

---

Do you think that a model which results in Biden getting 413 EVs as the most likely result is the same thing as a prediction that he will get 413 EVs?

---

Yes

---

No, that's not how it works.

413 may be the most probable result out of all the possibilities (the "mode"), but that is not to be interpreted as a "prediction" of that result. If we take the simulated results to be accurate, then you might predict 413 if the goal is to get it exactly right. But that is rarely the objective. Rather, the goal is to come as close as possible, which in these simulations is around 340-350 (near the "median").

quote:

---

otherwise there is no real purpose to doing a model of anything.

---

Yes there is, if you understand probabilities. Which of course most don't.

The only reason Silver posted this map is because of the unusual spike in the results at 413. You might expect more of a normal distribution, but due to the EC the results are kind of "spiky". No one should consider this to be a "prediction" or projection. Rather, it is just one possibility of many.

quote:

---

Of course they provide information.

It's kind of hard to persuade and/or manipulate without saying something.

---

I don't know, I see quite a few people persuaded by absolutely nothing at all.

quote:

---

When I use models in my business I am always trying to bring about a group consensus decision. Models are entirely adjustable in many directions.

---

The accuracy of the adjustments differentiates between information and mis/disinformation, while the intent of the adjustment (should it be inaccurate) differentiates between misinformation and disinformation.

Does this matter, in your view? Or is it all manipulation regardless?

quote:

---

I don't think biden gets over 400.

---

Me neither.

And neither does Nate Silver.

It is a 50/50 shot that Biden breaks 200 much less 400. He may get 400 months in the pen that is about it.

quote:

---

So....exactly what I said.

---

No

quote:

---

Kork, you come across as a really smart person and, despite my sophomoric bomb throwing post history, I'm making an honest effort to unpack what you are trying to say here.

Is accuracy dependent on belief or fact? I could absolutely believe that bad data is accurate and have pure intent. But I would also be misinforming by using it.

---

Accuracy depends on fact.

If a model is accurate, then it informs. If a model is not accurate, then it misinforms. Further, if it is intentionally inaccurate, then it disinforms.

Persuasion can of course be well-intentioned or deceitful. But you seem to be using "persuade" as a pejorative when it comes to models, as if they are always deceitful. I would call this "manipulation" rather than "persuasion".

So when I say that a model "informs", I mean that it is accurate and well-intentioned. And when you countered with basically "of course they persuade and/or manipulate with information", to me that's not information. It's disinformation.



In short, I've kind of lost track of what I'm trying to say here. I guess I just find it funny that any and every poll/story/whatever that is negative for Trump is immediately thought to be inaccurate and/or deceitful by most on this board.

quote:

---

I guess I just find it funny that any and every poll/story/whatever that is negative for Trump is immediately thought to be inaccurate and/or deceitful by most on this board.

---

They could be right,

this time.


But given the poor track record, they have to prove it.