For any, like myself, wondering "Who is Ben Welsh" ?
Hello. My name is Ben Welsh. I'm an Iowan living in New York City.
I am a reporter, an editor and a computer programmer. My job is to use those skills, together, to find and tell stories.
I work at Reuters, the world's largest multimedia news provider, where I founded the organization's News Applications Desk. In that role, I lead the development of dashboards, databases and automated systems that benefit clients, inform readers, empower reporters and serve the public interest.
[...]
I am certainly missing a lot of nuance here, but it seems to me Nate Silver managed to have his cake and eat it too. He surely got good money for selling FiveThirtyEight, and now that the buyer has erased the product, Nate can get back a huge chunk of its readers since he offers very similar analyses on his personal site. Sure, natesilver.net has less brand recognition than fivethirtyeight.com, but it's still decently well-known and can only go up from here.
I think the nuance is that it is notable historical articles about predictions and discussions of political elections, during a time when politics is quite at the fore-front of many people's minds
He may also finally shake off the comment trolls who piled onto him after 2016, seemingly blaming him for the election results (absurd but people are absurd).
After that election, a certain group would tirelessly work to discredit him any time his election predictions were not entirely one-sided.
It’s possible NS may have signed a contract saying that he cannot engage in elections prediction for X YZ months to same extent that he did with FiveThirtyEight.
> Electoral-vote.com is a website created by computer scientist Andrew S. Tanenbaum. The site's primary content was originally poll analysis to project election outcomes. Since the 2016 elections, the site also has featured daily commentary on political news stories.
I don’t really know much about it, but remember it as being _fantastic_ journalism every time I encountered one of their articles. As a bonus, great infographics and interactive data visualizations.
Unfortunately most of the most important visualizations are broken in the archived version. Including the gun deaths visualization and I think the P-hacking interactive
It's kinda sad to know no one else will get to experience those interactive visualizations. Though its nice to see the approval comparison page still works
Curious why they're broken, as the wayback machine seems to be able to run javascript. Do the visualisations rely on a server (or some other assets not included in wayback machine's crawl)?
It's not about robots.txt but yes, the owners of 538 can just send a cease and desist letter to get them all immediately removed. Many sites that don't want to preserve history have done this already.
This is a great service for everyone who appreciates thoughtful analyses about politics. Losing FiveThirtyEight was a big loss, but this archive helps. Bravo!
I think it's the fivethirtyeight of of historical significance, and Disney is one of the largest and wealthiest companies on the planet. So it's just kinda like "whoa, this is stratospheric negligence" or "whoa, what is the reason for this... assuming they are not idiots?"
Also, they don’t any plans for the IP, and Nate would’ve paid above-market rate just to take over and preserve the content for posterity. He estimates that they deleted 200,000 hours of human labor.
This is just some Disney suits being extraordinarily petty.
Here are some numbers roughly in the right ballpark: during the Disney era, which lasted about 10 years, FiveThirtyEight published about 20 stories a week. Let’s say that each story took about 20 hours to produce between research, writing, graphics and editing.3 Do the math, and that works out to about 200,000 person-hours of work that ABC News just deleted.
In a sense, nothing - and any other website should be archived, too.
In another sense, it's a journalistic source with information and commentary on past elections. Even aside from the political context that muddies the waters around or outright denies results, matters of public discourse on the web should not be ephemeral or subject to the decisions of the publication - they should be archived.
If I wanted to get the complete WARC archive of 538 - how do you do this in a friendly way? No interest in history tracking, just want the last available version from Internet Archive.
But that would be a false attribution. The Internet Archive did not create the index, Ben did. And the Internet Archive is not hosting the index, Ben is.
Ah, yes, could be worded better, fairplay. Point is the Ben attribution isn't needed in that place to avoid unnecessary confusion about who that is etc.
This is why people don't really buy the "but he had Trump at 30%, you just don't understand statistics" apologist line. Sure he hedged in the dying days of the campaign (a cynic might think to try to protect his credibility), but the tone overall was of a person who comprehensively failed to understand the mood of the country from beginning to end.
Which is a problem because these election predictions are not just pure "mathematical models" and "data driven" like 538 would have had you believe. What mathematical model should be used? What data should and should not be used? At some point those things are based on the modeller's understanding of reality.
I think Nate did a phenomenal job calling out pollsters in that time. Since 538 was predominately a poll aggregator that did tricky stats to rank the reliability of each poll. I remember specifically an interview with him griping about some of the unusual data he was seeing from pollsters that made it look like, and I quote, 'Someone has their finger on the scales'
Perhaps critiquing statistical methods used by polling was something he was good at. I have no real opinion of his work there, which I didn't pay attention to.
But predicting an election requires a lot more than polling datasets and statistics textbooks. That's the problem that he made himself out to be an election prediction wizard, but really that was off the back of his good prediction in quite a bland and conventional election.
When things got slightly more spicy and reality diverged from his vaunted "models", his "data science" predictably fell in a heap. The worst thing is almost not even that he got it wrong, it's that he seemed incapable of recognizing that present reality was quite significantly different from the past data he had used to build his models. Even after being wrong in so many of these predictions. He just kept churning out these pieces about how Trump was probably finished this time.
He didn’t hedge at the end. Nate always writes the models before election season then doesn’t touch them apart from actual bug fixes. The model actually organically predicted 30%.
I still think that’s about accurate. Maybe it should’ve been 40%.
Everyone forgets that it was a pretty close election. Clinton could’ve won without the Comey announcement.
I think he did hedge (or "strategically bug fix"). The prediction for Trump went from IIRC around 15 to 30 in the last week or so. It was a big swing, IIRC with a lot of waffle around why it happened but not a lot of verifiable fact.
> I still think that’s about accurate. Maybe it should’ve been 40%.
It wasn't accurate. This is something people misunderstand about these predictions. If the 2016 election was held 100 times, Trump would have won 100 times. It's not the same as rolling dice.
These election predictions don't say that. They say something like "the observations I have agree with scenarios that have Clinton winning, 70% of the time". Which is fine and correct as far as his data and model goes, but none of those scenarios were the reality he was trying to predict. They are all just figments of the model though. Getting down to the brass tacks, he predicted Clinton would win, and he was wrong.
Which is fine, we just can't know anything about his process from that failure. Certainly we can't conclude that it was "accurate", since it was not. If we had a good sample of elections where he used the same process and built up a good record then sure.
It's a bit more loaded than that. 538 post-Nate Silver had a model setup that was apparently kind of a mess. 538 was apparently sending strange messages to Republican leaning polling agencies, demanding they gave far more detailed audit information than usual (with the implication obviously being that they were fraudulent pollsters), and the guy running the site had fairly openly tuned his model on the assumption people cared about certain talking points. 538 was predicting Biden victories even when the polls were so overwhelmingly against him that not even the most Democratic leaning polling agencies had trust in him; even if you aren't running difficult math, that means something has gone wrong with the model.
(Something which got worse after Harris was picked, although every polling aggregator went barmy - there's suspicions that a lot of polling agencies aggressively normalized their data to avoid being seen as biased, leading to an almost 50/50 split.)
Biden got literally 0% of the vote despite 538 predicting him to win.
They chose to pretend Biden's mental decline wasn't happening because that was the Democrat party line at the time. How can you trust predictions from someone who is willing to manipulate his results to prop up his political party?
For any, like myself, wondering "Who is Ben Welsh" ?
~ https://palewi.re/who-is-ben-welsh/
Ben is one of my favorite people in the world of data journalism. He's the author of many excellent training courses in the field, including:
- https://github.com/palewire/first-python-notebook
- https://github.com/palewire/first-web-scraper
- https://github.com/palewire/first-graphics-app
(Submitted title was "Ben Welsh made an index of all FiveThirtyEight articles on the Internet Archive" - we've since changed it)
Cheers for the clarity, that'll help me look less weird wrt above comment to future historians of archived HN threads :-)
TBH I enjoyed looking up Ben and finding out what he's about and done in the past far more than I did just knowing there's a 538 archive on IA.
3 replies →
Can't believe Ben Welsh is not Welsh, and FiveThirtyEight has nothing to do with Wales
Don't click that link if you have any feelings of inadequacy about your achievements LOL
I am certainly missing a lot of nuance here, but it seems to me Nate Silver managed to have his cake and eat it too. He surely got good money for selling FiveThirtyEight, and now that the buyer has erased the product, Nate can get back a huge chunk of its readers since he offers very similar analyses on his personal site. Sure, natesilver.net has less brand recognition than fivethirtyeight.com, but it's still decently well-known and can only go up from here.
I think the nuance is that it is notable historical articles about predictions and discussions of political elections, during a time when politics is quite at the fore-front of many people's minds
He may also finally shake off the comment trolls who piled onto him after 2016, seemingly blaming him for the election results (absurd but people are absurd).
After that election, a certain group would tirelessly work to discredit him any time his election predictions were not entirely one-sided.
It was so funny, he mostly predicted the election correctly, but people confused probabilistic forecasting for sports lines.
It’s possible NS may have signed a contract saying that he cannot engage in elections prediction for X YZ months to same extent that he did with FiveThirtyEight.
He didn’t. He retained the IP of his election forecast models and he publishes the results from them on his newsletter, The Silver Bulletin
1 reply →
> Nate can get back a huge chunk of its readers
The downside is this furthers the divide between folks who pay for subscriptions and masses who get shoveled ad-powered slop.
https://electoral-vote.com is still up, running, and showing data. It has always been free and ad free.
https://en.wikipedia.org/wiki/Electoral-vote.com
> Electoral-vote.com is a website created by computer scientist Andrew S. Tanenbaum. The site's primary content was originally poll analysis to project election outcomes. Since the 2016 elections, the site also has featured daily commentary on political news stories.
He put up a substantial number of articles for free during the 2024 election, it felt less paywalled than much of the MSM these days.
1 reply →
Couldn't figure out why archiving FTE aricles matters, but a quick search yields:
> Thousands of FiveThirtyEight articles seemingly vanish from the internet
https://news.ycombinator.com/item?id=48197703
I don’t really know much about it, but remember it as being _fantastic_ journalism every time I encountered one of their articles. As a bonus, great infographics and interactive data visualizations.
Unfortunately most of the most important visualizations are broken in the archived version. Including the gun deaths visualization and I think the P-hacking interactive
https://web.archive.org/web/20230205124354/https://fivethirt...
It's kinda sad to know no one else will get to experience those interactive visualizations. Though its nice to see the approval comparison page still works
https://web.archive.org/web/20241031232233/https://projects....
Curious why they're broken, as the wayback machine seems to be able to run javascript. Do the visualisations rely on a server (or some other assets not included in wayback machine's crawl)?
I'm not a soccer guy, but I still think the piece on Lionel Messi was awesome
https://web.archive.org/web/20140701122958/http://fivethirty...
This is because whoever owns Fivethirtyeight now (ABC?) deleted the whole archive of articles on the site.
Don't we need more than an index of Archive.org because whomever controls the domain could robots.txt these out of existence if they wanted to?
It's not about robots.txt but yes, the owners of 538 can just send a cease and desist letter to get them all immediately removed. Many sites that don't want to preserve history have done this already.
Archive.org mostly ignores robots.txt
https://blog.archive.org/2017/04/17/robots-txt-meant-for-sea...
3 replies →
[flagged]
Please, say that again in comprehensible English.
1 reply →
This is a great service for everyone who appreciates thoughtful analyses about politics. Losing FiveThirtyEight was a big loss, but this archive helps. Bravo!
I'm seeing a lot about this. What makes this situation different than any other website going offline?
I think it's the fivethirtyeight of of historical significance, and Disney is one of the largest and wealthiest companies on the planet. So it's just kinda like "whoa, this is stratospheric negligence" or "whoa, what is the reason for this... assuming they are not idiots?"
Also, they don’t any plans for the IP, and Nate would’ve paid above-market rate just to take over and preserve the content for posterity. He estimates that they deleted 200,000 hours of human labor.
This is just some Disney suits being extraordinarily petty.
1 reply →
https://www.natesilver.net/p/disney-erased-fivethirtyeight
In a sense, nothing - and any other website should be archived, too.
In another sense, it's a journalistic source with information and commentary on past elections. Even aside from the political context that muddies the waters around or outright denies results, matters of public discourse on the web should not be ephemeral or subject to the decisions of the publication - they should be archived.
Related: Disney erased FiveThirtyEight
https://news.ycombinator.com/item?id=48197703
Github link:
https://github.com/palewire/fivethirtyeightindex.com
If I wanted to get the complete WARC archive of 538 - how do you do this in a friendly way? No interest in history tracking, just want the last available version from Internet Archive.
Love Ben but title can simply be: Index of FiveThirtyEight articles preserved by the Internet Archive
But that would be a false attribution. The Internet Archive did not create the index, Ben did. And the Internet Archive is not hosting the index, Ben is.
Ah, yes, could be worded better, fairplay. Point is the Ben attribution isn't needed in that place to avoid unnecessary confusion about who that is etc.
Those 2015-16 ones sure aged poorly, I'm reminded of this https://i.imgur.com/6Z9QQj3.jpeg
This is why people don't really buy the "but he had Trump at 30%, you just don't understand statistics" apologist line. Sure he hedged in the dying days of the campaign (a cynic might think to try to protect his credibility), but the tone overall was of a person who comprehensively failed to understand the mood of the country from beginning to end.
Which is a problem because these election predictions are not just pure "mathematical models" and "data driven" like 538 would have had you believe. What mathematical model should be used? What data should and should not be used? At some point those things are based on the modeller's understanding of reality.
I think Nate did a phenomenal job calling out pollsters in that time. Since 538 was predominately a poll aggregator that did tricky stats to rank the reliability of each poll. I remember specifically an interview with him griping about some of the unusual data he was seeing from pollsters that made it look like, and I quote, 'Someone has their finger on the scales'
Perhaps critiquing statistical methods used by polling was something he was good at. I have no real opinion of his work there, which I didn't pay attention to.
But predicting an election requires a lot more than polling datasets and statistics textbooks. That's the problem that he made himself out to be an election prediction wizard, but really that was off the back of his good prediction in quite a bland and conventional election.
When things got slightly more spicy and reality diverged from his vaunted "models", his "data science" predictably fell in a heap. The worst thing is almost not even that he got it wrong, it's that he seemed incapable of recognizing that present reality was quite significantly different from the past data he had used to build his models. Even after being wrong in so many of these predictions. He just kept churning out these pieces about how Trump was probably finished this time.
7 replies →
He didn’t hedge at the end. Nate always writes the models before election season then doesn’t touch them apart from actual bug fixes. The model actually organically predicted 30%.
I still think that’s about accurate. Maybe it should’ve been 40%.
Everyone forgets that it was a pretty close election. Clinton could’ve won without the Comey announcement.
I think he did hedge (or "strategically bug fix"). The prediction for Trump went from IIRC around 15 to 30 in the last week or so. It was a big swing, IIRC with a lot of waffle around why it happened but not a lot of verifiable fact.
> I still think that’s about accurate. Maybe it should’ve been 40%.
It wasn't accurate. This is something people misunderstand about these predictions. If the 2016 election was held 100 times, Trump would have won 100 times. It's not the same as rolling dice.
These election predictions don't say that. They say something like "the observations I have agree with scenarios that have Clinton winning, 70% of the time". Which is fine and correct as far as his data and model goes, but none of those scenarios were the reality he was trying to predict. They are all just figments of the model though. Getting down to the brass tacks, he predicted Clinton would win, and he was wrong.
Which is fine, we just can't know anything about his process from that failure. Certainly we can't conclude that it was "accurate", since it was not. If we had a good sample of elections where he used the same process and built up a good record then sure.
10 replies →
[dead]
[dead]
[flagged]
[dead]
[flagged]
Because of broad statistical illiteracy regarding early forecasts?
538: "We have no idea who will win. It seems like basically a coin toss."
Reality: someone wins.
Internet: "How did 538 mess this up so badly?"
It's a bit more loaded than that. 538 post-Nate Silver had a model setup that was apparently kind of a mess. 538 was apparently sending strange messages to Republican leaning polling agencies, demanding they gave far more detailed audit information than usual (with the implication obviously being that they were fraudulent pollsters), and the guy running the site had fairly openly tuned his model on the assumption people cared about certain talking points. 538 was predicting Biden victories even when the polls were so overwhelmingly against him that not even the most Democratic leaning polling agencies had trust in him; even if you aren't running difficult math, that means something has gone wrong with the model.
(Something which got worse after Harris was picked, although every polling aggregator went barmy - there's suspicions that a lot of polling agencies aggressively normalized their data to avoid being seen as biased, leading to an almost 50/50 split.)
1 reply →
Biden got literally 0% of the vote despite 538 predicting him to win.
They chose to pretend Biden's mental decline wasn't happening because that was the Democrat party line at the time. How can you trust predictions from someone who is willing to manipulate his results to prop up his political party?
1 reply →
... so an event that had a (simulated) 47% percent chance of happening happens, and an event that had a 53% chance of happening didn't.
You must _hate_ it when you need <= 4 and you roll a 5...
They had Trump 1/3 and Clinton 2/3 IIRC. That was a bigger spread. I remember Nate Silver saying that is what 1/3 probability looks like later.