Is it SLOP??

 

I came upon this video the other day, and found it quite interesting. It reminded me of a post I wrote on here nearly a year ago in response to this video on The Athletic's Tifo sub-channel, discussing why football is such an exceptionally complex game (almost uniquely so), and why this imposes severe limits on the extent to which data analysis can be helpful in 'understanding' it. This new video is a far more superficial discussion of the topic than the earlier Tifo one, focusing mainly on why football ('soccer') is so much more 'unpredictable' than the major American team sports. Unsurprisingly, it's because it's a low-scoring game (so, a single error leading to a goal can more often have a decisive impact on the final result), and because a draw being included among the possible game outcomes heavily impacts both the game's tactics and the predictability of results. The one major piece of data analysis in the video suggests that from 2005-2025 barely half of Premier League games were actually won by the 'favourite' (although that begs the question of how the 'favourite' is assessed; many fixtures are so tight that there is no clear favourite - certainly not when other factors like recent form and home advantage are taken into consideration; it also omits to consider how many of the 'upset' results were only draws rather than losses, which obstructs direct comparability with the American team sports).

While I quite liked this video, I do feel slightly hesitant about sharing it. Youtube is awash with AI SLOP these days, and I'm not completely confident that this isn't another example. It does seem to be free of any of the usual tell-tales - the heavy-handed rhetorical antitheses, the frequent repetition of content, or occasional obvious glitches in the voiceover (AI-generated narration tends to have the odd clunking mispronunciation or bizarre bit of phrasing or intonation now and then, or sometimes just an obvious break in the continuity mid-line - which completely gives it away). And the content appears to be all accurate and true (I haven't checked that central statistical claim, though....). But it is a bit glib and shallow. And they are churning out an awful lot of content in a very short time: over two dozen videos in barely two months since the channel launched. Alas, I think I smell a rat. But I'll probably check out a few more of their videos to try to find some persuasive evidence for my hunch on this.