Tannehill and the Offensive Line

[fyo]

Dolphins:  324 passing plays, 44 (14%) with 2 or 3 TE sets; 244 (75%) with a lone setback; 67 (21%) with 4 WRs.
Cowboys:  332 passing plays, 38 (11%) with 2 or 3 TE sets; 216 (65%) with a lone setback; 107 (32%) with 4 WRs.

If it weren't inappropriate, I'd be rolling around grasping my sides. Or throwing up. Or just giving up.

Seriously, I'm not going to do this 10 times every single week. Go watch some football. Have you heard of a guy called Jason Witten? He's a tight end for the Dallas Cowboys. He has 49 receptions this year (more than 50% more than all Dolphins TEs combined). Do you think it's possible tight ends don't always perform the same function in an offense? How often do you think Witten stays in to block compared to Gray / Sims / Jones? That is the relevant statistic, not the number of tight ends on the field in this age of the receiving tight end.

[Dolfanalyst]

...you said in response to an article attempting to provide proof that Tannehill's supporting cast is below average.

This is why I don't bother putting much effort into these responses to you.  I see the amount of effort that FO put into their article (and fyo put into his analysis) and you come back with this "Yeah, but they're probably still average anyway so that article is wrong" nonsense without even a pretense of trying to support that claim.

The problem is that you're placing credence in the article's position without holding it to an appropriate standard of supporting its own claim.

Claims aren't supported by "effort apparent to Spider-Dan."  They're supported by the appropriate rigor with regard to the material.

[Dolfanalyst]

If it weren't inappropriate, I'd be rolling around grasping my sides. Or throwing up. Or just giving up.

Seriously, I'm not going to do this 10 times every single week. Go watch some football. Have you heard of a guy called Jason Witten? He's a tight end for the Dallas Cowboys. He has 49 receptions this year (more than 50% more than all Dolphins TEs combined). Do you think it's possible tight ends don't always perform the same function in an offense? How often do you think Witten stays in to block compared to Gray / Sims / Jones? That is the relevant statistic, not the number of tight ends on the field in this age of the receiving tight end.

If you tell me I might be persuaded.

Please, tell me, and allow me to verify that with something more objective and definitive than "go watch a game" or "because I said so."

[fyo]

If you tell me I might be persuaded.

Please, tell me, and allow me to verify that with something more objective and definitive than "go watch a game" or "because I said so."

You're the one saying the sky is green. Your claim that offensive lines are basically all the same, with very little variation, and thus no net effect on the outcome of games is beyond preposterous.

[Spider-Dan]

Dolfanalyst, if you think my responses have lacked effort so far, that was before I knew that you seriously consider "I suspect that if we dig deeper, we will find that the players in question are about average" a devastatingly effective rebuttal that simultaneously 1) requires no further evidence and 2) swiftly trivializes opposing evidence given.

I thought I caught you in a hole in your logic, but it turns out that that's your actual position.

[Dolfanalyst]

You're the one saying the sky is green. Your claim that offensive lines are basically all the same, with very little variation, and thus no net effect on the outcome of games is beyond preposterous.

And once again, if you tell me why, and provide something to support your position that's more objective and definitive than "go watch a game" or "because I said so," I'll very likely be persuaded!

I'm very open-minded, but only to a certain kind of evidence.  Ipse dixit ain't it.

[Dolfanalyst]

Dolfanalyst, if you think my responses have lacked effort so far, that was before I knew that you seriously consider "I suspect that if we dig deeper, we will find that the players in question are about average" a devastatingly effective rebuttal that simultaneously 1) requires no further evidence and 2) swiftly trivializes opposing evidence given.

I thought I caught you in a hole in your logic, but it turns out that that's your actual position.

It trivializes the evidence given only if that evidence is so superficial as to permit no rejection of the null hypothesis (i.e., things are likely average until proven otherwise).

Again, I'm open-minded, but I'm not gullible.

[Dolfanalyst]

You're the one saying the sky is green. Your claim that offensive lines are basically all the same, with very little variation, and thus no net effect on the outcome of games is beyond preposterous.

Also, and I've said this before, it isn't preposterous on theoretical grounds to believe that the parity in the league causes the largest unit on the field (five players) to have comparatively little variation from team to team.

The more players in a unit, the greater the likelihood that the parity the league is founded on will lessen the variation in those units across the league.  It's hardly "preposterous" to believe that.

Again, you seem to be arguing from the position of "offensive line versus no offensive line," in terms of how that would affect a team, rather than from the position that every team in fact does have an offensive line, and so the question becomes how much they differ from each other, not from some hypothetical world in which there is no offensive line at all.

Yes, it's indeed preposterous to think that a team with no offensive line at all would perform just as well as a team that had one.  But of course that isn't what I'm saying here.

[Spider-Dan]

No remotely reasonable person would think fyo is arguing for a case of 6-on-11 football, where the football somehow magically snaps itself.
But I guess it's easier to attack that strawman than to continue to defend your ridiculous position that the performance of the offensive line doesn't matter.

[Dolfanalyst]

No remotely reasonable person would think fyo is arguing for a case of 6-on-11 football, where the football somehow magically snaps itself.
But I guess it's easier to attack that strawman than to continue to defend your ridiculous position that the performance of the offensive live doesn't matter.

I'm simply trying to make sense of his statement that it's "preposterous" to think there is comparatively little variation among offensive lines throughout the league, and that's all I can come up with, that he must be misinterpreting my position to mean "offensive line versus no offensive line."

Just trying to reach some common ground, if possible, that's all.

Ironically though, within your post above, you've again caricatured my position (i.e., the same straw man you're speaking of).

I haven't said "the performance of the offensive line doesn't matter."

Once again, what I have said is that offensive lines, according to the quantitative measures of them we have available, differ so little from each other that they can't possibly explain the differences we see in QB play.

When QBs differ from each other at (hypothetical number) 100, and offensive lines differ from each other at only (hypothetical number) 10, offensive line play can't possibly explain QB play to a significant degree.

The variation in QB play has to be explained by something else.

[Spider-Dan]

You accuse me of constructing a strawman when I summarize your position as "the offensive line doesn't matter."  Fine.

You literally just said that the offensive line is an insufficient explanation for level of QB play (i.e. it doesn't matter).  Again, fine.  So what, exactly, does the offensive line matter for... the run game?  That can't be true, because (unless you have a new groundbreaking theory to unveil) the effectiveness of the run game definitely affects the passing game (i.e. the play of the QB).

[Dolfanalyst]

You accuse me of constructing a strawman when I summarize your position as "the offensive line doesn't matter."  Fine.

You literally just said that the offensive line is an insufficient explanation for level of QB play (i.e. it doesn't matter).  Again, fine.  So what, exactly, does the offensive line matter for... the run game?  That can't be true, because (unless you have a new groundbreaking theory to unveil) the effectiveness of the run game definitely affects the passing game (i.e. the play of the QB).

The degree to which it matters is a direct reflection of the correlation between offensive line play and QB play league-wide.

Consider the following hypothetical correlations:

Between offensive line play and QB play -- 0.25
Between running game play and QB play -- 0.15
Between receiver play and QB play -- 0.35
Between QB individual ability and QB play -- 0.90
Between luck (random variables) and QB play -- 0.15

All of those contributions to QB play "matter," but clearly one of them matters far beyond all the others, and we should therefore attribute differences in QB play to that one well before doing so with the others.  That one should be our "default" explanation for QB play, not the offensive line.

My point here is that the variation in offensive line play across the league is consistent with those numbers above, in that it suggests that offensive lines have such a minimal relationship with QB play.

[Spider-Dan]

I could also make up some hypothetical numbers that support the point I'm trying to make, but that seems like a complete waste of time.

[Dolfanalyst]

I could also make up some hypothetical numbers that support the point I'm trying to make, but that seems like a complete waste of time.

The problem is that every time someone has tried to take the position that the offensive line affects Tannehill (or QBs in general) "yay" much, and they nominate some way of measuring offensive line play in that regard, the correlation ends up being roughly what I noted above -- 0.25, which is relatively weak.  That number in fact isn't hypothetical.

So, the issue is precisely the opposite of the way you framed it in the quote above.  Other people are proposing a relationship, and the objective data don't support their position.  My position on the other hand is supported by the objective data!

[Dolfanalyst]

Who would've predicted that Tannehill would've been sacked only twice and posted a 130.6 QB rating and 9.5 YPA with Albert, Tunsil, and Pouncey inactive?