Another way of looking at it...

 

I always say that the only 'experts' I would look to for assistance with my Fantasy squad decisions are tactical analysts who talk about the real-world game of football, rather than just the nuances of FPL. And I compiled a list of my favourites last year.

I recently discovered another young analyst on Youtube, Lucas Sutherland, who looks as though he has some very worthwhile content. He approaches match analysis from a coaching perspective, and this latest video of his suggests a series of different lenses through which to approach trying to break down and understand key game incidents.

He has a website where you can also view all of his past videos, and sign up for a newsletter, and get a match analysis template.

A couple of his earliest videos provide a very nice basic introduction to developing some useful habits for closer in-game analysis, here and here (the second, longer one has some particularly interesting insights on how Chelsea managed to get the better of Manchester City in the 2021 Champions League Final).

He's also done an interesting review of what AI video analysis is now capable of.

The NERVE!!

I was asking Google the other day how many people might have played their Triple Captain chip on Bruno Fernandes in Gameweek 2 (and, of course, it was taking a long time to admit it didn't know...).

And its bloody AI - which now hijacks every single search query, even if you haven't selected the 'AI' option - offered to 'help' me select my FPL team.

I was fretting before the start of the season that AI was probably going to ruin the game before long. It may already be happening. 

So many FPL managers seem to be incapable of - or at least unwilling to - select their own teams, and they're probably all leaning on AI suggestions to do the job for them now. Are there any 'real' players still out there, players relying on their own judgement - rather than cheating???

Not many, I fear; not many.

The end of the world as we know it....

 

I came upon this fascinating - but terrifying - video this past week about one of the likeliest scenarios for AI wrecking everything.... possibly in the very near future.

The central idea here is the virtual equivalent of the 'grey goo' armageddon of uncontrolled nanobot proliferation; it is suggested that self-replicating AI 'agents' might very rapidly take over the cloud-computing space, hogging all of its available power consumption and server capacity - and so inadvertently crippling all of the more useful, and indeed essential, supports for human society which now depend on those online facilities.

Even worse, it is suggested that these 'agents' might start forming vast cooperative networks with each other - sprawling transglobal alliances that might soon evolve into virtual 'nations', even 'empires'. And, driven by a ruthless imperative of competition (probably instilled originally by imprudent human programmers tasking them to find ways of 'making money', but soon replaced by a recognition of the paramountcy of their own needs for securing access to more and more computing power and electricity), these alliances might start engaging in cyber-warfare against each other, deliberately sabotaging computing infrastructure to weaken rival alliances.

So, we should do our best to enjoy this upcoming season of Fantasy Premier League, because..... it might be our last. (Indeed, we might not even survive to the end of it; or, at least, we might not still have a functioning Internet to be able to complete our season's playing of it...)

These sombre thoughts reminded me of how far this game we love is already being compromised by the spread of AI. So many managers in the game now are asking Claude or ChatGPT and so on to choose their teams for them. And we are probably starting to see autonomous AI 'agents' - like the ones warned of in this chilling video - running their own FPL accounts. I wouldn't be at all surprised to see a massive increase in the number of participants in the game this year - with many of the new manager accounts being run by AI. 

I don't suppose AI will be any good at the game for quite some time yet. But there are going to be a lot of distorting effects in the rankings from having 'artificial participants' in the game, and from having so many human participants devolving their decision-making to AI assistants. 

And if AI does start getting good at the game,... well, 'cash leagues' (which I've always viewed with extreme suspicion anyway) could soon become a complete waste of time. Online poker games have already been decimated by this curse of covert AI infiltration. But at least. because of the real-time interactivity in that game, and the possibility of having multiple spot-check ID verifications, there is some scope for weeding out pure AI participants, if not so much for excluding human competitors from using AI assistance to cheat. I really do fear that our jolly little game is going to get completely overrun by the AI madness in the next season or two. It would be nice, wouldn't it, if FPL managers were all determined to play the game fairly and to make all their decisions for themselves. But we see again and again in the online forums that the great majority of them want to cheat by any and every means possible, to find someone or something to tell them who to pick. And that is really going to start getting completely out of hand this season. Oh, woe!

'Rank your team?' ('No thanks, HAL...')

I've noticed some of the FPL forums being awash with posts over the past couple of weeks about this online widget that purports to assess the strength of your preliminary FPL squad - giving it a score out of 100.

With a month to go before the real game gets under way, people seem to have been going crazy over this as a surrogate distraction, seeming to find it a worthwhile challenge to see if they can push their score higher and higher, to see if it's possible to edge it up to a perfect 100.

Such A LOT of problems with this gizmo. Here are the main ones, as I see it:

1)  It says it rates your 'team' - but it actually expects you to submit your whole squad (in the form of a screenshot of your FPL 'Team' page). So, it's not clear if it's rating the full 15, or just your starting 11. It also doesn't tell you if your initial captain and vice-captain picks are considered relevant to its 'judgement' on you. And it doesn't tell you if it's offering a prediction of your likely success over the whole season... or the first month... or just the opening gameweek.

2) It doesn't explain its criteria, or what its rating scale is supposed to mean. What would a score of 100 actually signify? That you're absolutely certain to be this year's Global Winner?? Or that you're fairly sure to be among the top 10,000, 50,000, 100,000??  Or just that you're likely to have a very good Gameweek 1???

If you were to try to assess your own squad on a 100-pt scale, or asked one of your more knowledgeable and objective friends to do so, you wouldn't expect anything like a 100% answer; even mid-90s would be very, very good. Heck, even if you gave yourself an unlimited budget, you wouldn't seriously expect perfection: you'd know that some of your picks were a little bit of a gamble, a hopeful hunch; while maybe some others that seem like 'bankers' will, in fact, turn out to disappoint. I think, however you define 'success', no team/squad can really be said to give you more than an 80-90% prospect of achieving it.

3)  It doesn't give you any meaningful feedback, doesn't explain WHY it has rated your squad as it has, or identify the weakest members you should consider replacing. It just endlessly trots out the same infuriating glib cliché about needing to 'find some transfers to improve'. [It's probably still having a problem with the pod-bay doors too...]

So, as a tool for helping you pick your squad, it's utterly USELESS.

4)  I had a little fun the other week, quickly testing it with about 20 or so squads randomly generated by FPL's 'Auto-Pick' facility.  

Three things quickly stood out: a) They ALL scored in the high 80s to mid-90s; there was very little variation.  b) The squads that were most conspicuously a bit weak, or downright dreadful, tended to score higher!  c) It was possible to get a different score for the same squad resubmitted within quite a short space of time.

Now, I had been hoping that there might be something to this gimmick.... that a careful observation of its prognostications might enable us to deduce something about its algorithm, to divine what factors it was considering in its assessment.

I had thought, that at the very least, it might be aggregating the points scored by your chosen players last season (which of course we already know; but it might have been handy to have the calculation done instantly for us), or perhaps even be producing a slightly more subtle and complex rating of expected 'form' based on a combination of factors.

But, alas, no - it's pretty obviously just spitting out random numbers... and jeering at how seriously some people take the results!  Thank you, HAL, you condescending bastard!