After watching the full video and reading the comments exchanged, I don't believe Veritasium there is guilty of anything more than believing studies with questionable funding sources and buying into autonomous cars too soon.
Tom Nicholas' set of arguments are in general good fodder for discussion but the premise is ultimately undermined by the video clickbait title and thumbnail calling several different creators' works abject propaganda. I get it, there's a counterculture cottage industry to be had going after larger creators.
To pick one example of problematic information: Tom presents his argument that the figure of 94% of accidents being caused by human error is a misleading statistic and argues an example scenario that would seem to assign blame to the driver unfairly (a hedge or tree branch obscures a stop sign, road markers have been worn away, etc.).
The issue with this presentation is that as a counterargument to the roughly 2.1 million accidents being compared to, it utterly fails on its face. There's no possible way that represents a substantial enough portion of the accidents to be more than a rounding error. I haven't poured over the NHTSA/CDC data around that, obviously but this just doesn't pass the smell test of "how likely is the described counter-scenario?" However, the overwhelming majority of accidents have been shown to occur close to home during a daily commute, in areas you are likely familiar with.[1] Additionally, it also doesn't square with the evidence that shows the majority (~55%) of fatal car accidents are single vehicle accidents.[2] It's an example of a contrived example that is so contrived so as to be meaningless. In a section of the video with the title "Lying with Statistics" it's an interesting choice to Lie without Statistics instead.
If you don't believe the researchers of the Waymo-funded study were ethical based wholly on where their money comes from and its lack of peer review, that's very reasonable. However, the funding source is less problematic to me than the lack of peer review because the system of scientific study is broken throughout the world, and we often see companies commissioning studies for whatever their particular area of operation is because they're the only ones with a vested interest in examining it. I'd love to see a separate solution for that.
Another example: In his text responses to Veritasium, Tom responds with:
It is, again, completely disingenuous to refuse to mention the maps which Waymo vehicles rely on when it suits your argument during your initial video but to now hold them up as a vital part of the technology. You spoke as though those maps didn’t exist in the video and so it was only fair, in critiquing your video, to respond in kind.
A further point for consideration here is the extent to which creating maps/scans for the entirety of even the United States (let alone other countries too) is practical. That seems like a highly intensive task which would be very costly, possibly to the point it’s unworkable.
I would hope the flaw in this concept is obvious to anyone here but there are OPEN SOURCE high-quality sources of mapping information, let alone maps that could be built by someone with funding.
Secondly, the idea of crying foul that because Veritasium didn't mention the maps "enough" in their video when they they responded to him about it is a further example of goalpost shifting. Furthermore, if you're trying to get to the source of truth, don't you think it's wise in terms of critique to you know, BE ACCURATE about what you're saying? This kind of thing should be unacceptable for someone proclaiming to be exposing "the extent to which a creator signing up to one of these sponsorship deals results in them compromising the editorial content of their videos to the point where education becomes misinformation."
I'm not going to rehash the entire argument between them. I think the primary thing Veritasium is guilty of here is of overstating and over-believing how ready for use autonomous cars are. I get it, Derick's ridden in one, so that likely leaves an impression (the psychology of "I wouldn't have done this if it was unsafe" at play). I disagree though, as I've said in other threads on HN, I'm staunchly against putting these things on the road until they're better than humans. It certainly doesn't rise to the level of propaganda in my opinion, even though I disagree with Veritasium on how ready the things are.
This is a good example of why I'd recommend the channel though. Veritasium actually responded to the criticism video with their own points and responded to Tom Nicholas when he sent them requests for comment. If nothing else this at least demonstrates a willingness to engage.
> After watching the full video and reading the comments exchanged, I don't believe Veritasium there is guilty of anything more than believing studies with questionable funding sources and buying into autonomous cars too soon.
Perhaps. But even if that was the case, the circumstances around it was enough to destroy my trust for Veritasium.
The way in which Waymo was interacting with him and other YouTubers should have set off all kinds of alarm bells. I couldn't come up with a more perfect hypothetical situation in which I'd expect him to be fully engaged in skepticism and critical analysis.
I normally wouldn't blame people too much for any specific instance of getting caught up in something cool. But if there was ever a time to do so, it was this.
A big company in the most recent field joining the list of perpetually-just-around-the-corner technology, and one aimed directly at consumers at that, grabs a bunch of YouTubers together to make a bunch of videos for simultaneous(-ish?) release and gives them some group presentations and special access to the company. If you aren't going to be maximally critical of the companies claims in this circumstance, when are you?
So even if he just got caught up in the coolness despite the best intentions, I don't think he deserves a pass. I can't just go back to assuming he's actually properly evaluating his sources in all his other videos.
And it wasn't even like we just got a video devoid of heavy review. We got a video with outright non-cited nor disclaimed statements straight off the presser being presented as if they were his own original thoughts/analysis.
The real problem for me here was mostly one of ethics and trust. I didn't feel like the way he went about making the video was particularly ethical - especially without a much stronger level of disclaimer. And I lost my trust in his ability to actually critically review the information he presented.
If I'm going to have to watch his videos with full alertness and skepticism, his videos aren't useful to me. I don't watch Veritasium or other such channels for information that is all that important to me, but rather as a mildly intellectual way to pass time and be exposed to new ideas - often while eating, doing something menial, or winding down for the day. It's still important to me that that information is reasonably accurate, as it still tends to join the hoard of random things in my head that 'I've heard somewhere...'.
If I don't trust the presenter to both have a good head on their shoulders, and to have actually used it in good faith when creating the video, their videos are useless to me.
> I would hope the flaw in this concept is obvious to anyone here but there are OPEN SOURCE high-quality sources of mapping information, let alone maps that could be built by someone with funding.
I honestly don't remember much about this specific point by now, but the use of the word 'scans' in the quote makes me suspect we're talking not about just about something like OSM but rather something more like a curated and processed point-cloud.
If so, that does seem like something beyond the capabilities of the open-source mapping movement without crazy amounts of funding just to generate the raw point clouds. And even then, how many volunteers do you need to validate a map/scan as correct and accurate enough for a self-driving car to use it as a primary data source?
But maybe I'm just getting thrown off by the word 'scan' and this is a non-issue.
As for the rest... honestly I'd have to dive in to even begin to address any of it. I don't doubt Tom Nicholas made some errors, at least. But as I said earlier, this stuff wasn't really my main concern to begin with.
> This is a good example of why I'd recommend the channel though. Veritasium actually responded to the criticism video with their own points
Conversely, this was one of the major strikes for me. Of course responding is good in general, but I found his response extremely poor. Based on memory, I felt like he avoided the larger issues and instead quibbled about specific details that weren't even related to the major issues I had. I was left thinking that he was clearly more interested in PR than in a good-faith discussion about the various issues.
To be fair, at the point I read the response, I was already disinclined to cut Veritasium a whole lot of slack. But his response was about as bad as it got for me, and cut off any desire I had to continue giving him some slack.
> Tom Nicholas' set of arguments are in general good fodder for discussion but the premise is ultimately undermined by the video clickbait title and thumbnail calling several different creators' works abject propaganda.
I won't disagree with you there. I'd honestly forgotten about that aspect of his video by now. Though if you're willing to cut Veritasium some slack on his titles and thumbnails, I'd say Tom Nicholas deserves some too. But really I'd prefer nobody did it in the first place.
I'm guessing his presentation has similar issues even though you didn't mention it. And I agree with you again. I've actually just stopped watching some otherwise interesting videos partway through recently because that sort of thing started bugging me so much. I can't fault you there.
Tom Nicholas' set of arguments are in general good fodder for discussion but the premise is ultimately undermined by the video clickbait title and thumbnail calling several different creators' works abject propaganda. I get it, there's a counterculture cottage industry to be had going after larger creators.
To pick one example of problematic information: Tom presents his argument that the figure of 94% of accidents being caused by human error is a misleading statistic and argues an example scenario that would seem to assign blame to the driver unfairly (a hedge or tree branch obscures a stop sign, road markers have been worn away, etc.).
The issue with this presentation is that as a counterargument to the roughly 2.1 million accidents being compared to, it utterly fails on its face. There's no possible way that represents a substantial enough portion of the accidents to be more than a rounding error. I haven't poured over the NHTSA/CDC data around that, obviously but this just doesn't pass the smell test of "how likely is the described counter-scenario?" However, the overwhelming majority of accidents have been shown to occur close to home during a daily commute, in areas you are likely familiar with.[1] Additionally, it also doesn't square with the evidence that shows the majority (~55%) of fatal car accidents are single vehicle accidents.[2] It's an example of a contrived example that is so contrived so as to be meaningless. In a section of the video with the title "Lying with Statistics" it's an interesting choice to Lie without Statistics instead.
If you don't believe the researchers of the Waymo-funded study were ethical based wholly on where their money comes from and its lack of peer review, that's very reasonable. However, the funding source is less problematic to me than the lack of peer review because the system of scientific study is broken throughout the world, and we often see companies commissioning studies for whatever their particular area of operation is because they're the only ones with a vested interest in examining it. I'd love to see a separate solution for that.
Another example: In his text responses to Veritasium, Tom responds with:
I would hope the flaw in this concept is obvious to anyone here but there are OPEN SOURCE high-quality sources of mapping information, let alone maps that could be built by someone with funding.Secondly, the idea of crying foul that because Veritasium didn't mention the maps "enough" in their video when they they responded to him about it is a further example of goalpost shifting. Furthermore, if you're trying to get to the source of truth, don't you think it's wise in terms of critique to you know, BE ACCURATE about what you're saying? This kind of thing should be unacceptable for someone proclaiming to be exposing "the extent to which a creator signing up to one of these sponsorship deals results in them compromising the editorial content of their videos to the point where education becomes misinformation."
I'm not going to rehash the entire argument between them. I think the primary thing Veritasium is guilty of here is of overstating and over-believing how ready for use autonomous cars are. I get it, Derick's ridden in one, so that likely leaves an impression (the psychology of "I wouldn't have done this if it was unsafe" at play). I disagree though, as I've said in other threads on HN, I'm staunchly against putting these things on the road until they're better than humans. It certainly doesn't rise to the level of propaganda in my opinion, even though I disagree with Veritasium on how ready the things are.
This is a good example of why I'd recommend the channel though. Veritasium actually responded to the criticism video with their own points and responded to Tom Nicholas when he sent them requests for comment. If nothing else this at least demonstrates a willingness to engage.
[1]: https://www.autoweek.com/news/a2108966/survey-finds-vehicle-... (I can only find contemporary articles, not the survey itself, though this particular fact is repeated here in Australia/New Zealand with an actual study: https://www.scimex.org/newsfeed/yes,-car-crashes-are-more-li...)
[2]: https://www.iihs.org/topics/fatality-statistics/detail/urban...