←back to thread

CDC gets list of forbidden words

(www.washingtonpost.com)
382 points js2 | 9 comments | 16 Dec 17 02:17 UTC | HN request time: 1.323s | source | bottom
Show context
jimjimjim ◴[16 Dec 17 03:29 UTC] No.15937289[source]▶
How can anybody actually support the current administration if this sort of thing is what is being pushed. how is this improving the world? how is this better? how is this good?

anyone?

replies(3): >>15937336 #>>15937379 #>>15937472 #
ryanx435 ◴[16 Dec 17 03:55 UTC] No.15937472[source]▶
If I answer truthfully I'm going to be down voted and shamed, so I won't answer.

Instead, I'm going to continue to be part of the silent majority that supports him, but doesn't speak out because we are often falsely labeled as deplorables or worse.

Edit: people like me are what you get when society stifles free speech: unwilling to engage and so continuing in our beliefs silently. maybe you coastal elitists should stop shaming people for having different opinions and mental models of how the world works.

And if you dummies think that speech isn't being stifled, look no further than James Demore being fired for trying to figure out root cause analysis of why there aren't more women in tech

Edit 2: someone is going through my post history and downvoting. Good job, ladies, you're proving my point.

Edit 3: and now I'm not allowed to respond to people because I'm posting "too fast". Great.

replies(8): >>15937504 #>>15937523 #>>15937527 #>>15937723 #>>15937759 #>>15937846 #>>15937852 #>>15938297 #
scrollaway ◴[16 Dec 17 04:01 UTC] No.15937523[source]▶
You're going to be downvoted either way, because a post that just complains about theoretical downvotes gets downvoted on this site. So you might as well take the opportunity to explain.

It's also worth mentioning that the question GP is asking isn't really "how can people support the administration", as there's many, many answers to it. The real question is "how can people support this?".

And frankly, enlighten me. Do you support this?

Edit: I should also remind you that Trump is not supported by a "silent majority". Quite the opposite: Trump is supported and was elected by a vocal minority. His approval rating today is the lowest it's ever been according to 538's aggregations. It's time americans get their head out of the sand on this topic.

replies(1): >>15937804 #
1. scrollaway ◴[16 Dec 17 05:09 UTC] No.15937843[source]▶
1. Why are you not answering my question?

2. Why are you saying something patently false about 538? They gave Trump the highest chances to win across most polls pre-election. They wrote several articles about how Trump could still win despite the polls.

3. Why are you equating election predictive polling with aggregation of approval ratings?

4. What's with your hostile tone?

You're a very unpleasant man, aren't you, calling people "you dummies", being overtly aggressive and playing the victim? Maybe it's time for you to face that you're being downvoted not because you "have different opinions and mental models of how the world work", but simply because you're a nasty person. After all, you haven't actually expressed any opinion yet, and you're still being downvoted, so it can't be because of your opinions.

Here's something for you to think about: I have different opinions and mental models of how the world works. I often enough say controversial shit, and yet I manage to do so mostly without getting flagged. Clearly, your approach is wrong.

replies(1): >>15938294 #
2. ceejayoz ◴[16 Dec 17 05:20 UTC] No.15937875[source]▶
538 doesn't do polling. They aggregate and analyze others' polls.

538 gave Trump a ~30% chance of winning, where virtually all other outlets (Princeton, NYT, HuffPo, etc.) were giving him a 1-5%. They appear to have had the best handle on Trump's chances.

replies(1): >>15938285 #
3. dragonwriter ◴[16 Dec 17 05:35 UTC] No.15937932[source]▶
> They though Hillary was gonna win in a landslide.

Actually, 538 was pointing out why other poll-based predictors were overrating her likelihood of winning (because, unlike 538 and contrary to historical evidence, thet were largely assuming deviation from polling averages in the election would be independent between different states rather than correlated.)

4. smsm42 ◴[16 Dec 17 07:14 UTC] No.15938285[source]▶
That illustrates how useless such predictions are. What "Trump has 30% chance of winning" means? If Trump won, did this prediction work or not? If he lost, did it work or not? How one could distinguish successful prediction from a failed one, even post-factum? I don't see any way. Now, if we had 100 identical Trumps running in 100 USA elections, and 30 of them would win and 70 of them would lose, we'd say good work 538, you were spot on. But since we have one Trump winning one election - I do not see any way of seeing how 538 was "better" than NYT since I can't see how you can say whether both were successful or not. NYT gave Trump 1% to win, he won - so that 1% worked. Big deal, people win lottery all the time, with much lower chances. So maybe NYT was completely right and 538 is full of it, maybe it's really an one in a 100 event? How would you know?
replies(2): >>15941639 #>>15949068 #
5. dang ◴[16 Dec 17 07:16 UTC] No.15938294[source]▶
As you well know, you can't cross into this kind of personal attack here. I don't want to ban you, so please fix this and stop doing it, regardless of how wrong or unpleasant the other person is.

https://news.ycombinator.com/newsguidelines.html

6. ceejayoz ◴[16 Dec 17 20:00 UTC] No.15941639{3}[source]▶
We know 538's model was better because of what happened on election day.

Their model gave Trump a dramatically higher chance than other poll aggregators for a variety of reasons (detailed at http://fivethirtyeight.com/features/election-update-why-our-...) and #4 there (correlated polling errors between states) wound up being very significant.

Now, you're right that we can't perfectly determine if 99% or 70% (or something else entirely) was Hillary's actual chance of winning. We can, however, look at the assumptions made by each model and evaluate those on their merits.

7. dllthomas ◴[18 Dec 17 02:59 UTC] No.15949068{3}[source]▶
Over many predictions we can apply a proper scoring rule to judge calibration. It's certainly the case that we don't learn much from looking at a single prediction.

https://en.wikipedia.org/wiki/Scoring_rule#Proper_scoring_ru...

replies(1): >>15982738 #
8. smsm42 ◴[21 Dec 17 21:04 UTC] No.15982738{4}[source]▶
Sure, if we have many predictions, that makes sense. But there are not that many presidential elections, even less - ones that 538 predicted on, and even less - ones that 538 predicted on used the same methodology. So how does one evaluate?
replies(1): >>15983650 #
9. dllthomas ◴[21 Dec 17 22:55 UTC] No.15983650{5}[source]▶
If we have fewer predictions, we need to be correspondingly less confident in our assessment of the quality of the predictor, to be sure. 538 has made hundreds (at least) of predictions, and across those has seemingly performed well. You can quite reasonably make a claim that most of those don't translate perfectly to the specific setting of a presidential election. But at the worst we should treat them as irrelevant and conclude that we don't know much about 538's quality. That's a far cry from a demonstration that their predictions are poor, which is what was claimed.