constraint theory

900. The video from the boards:

Can we start calling the team Hokemen, Daily?

Speaking of…

We have fewer lame-o wins than anyone! Except Georgia! Pat Forde scours the ten winningest programs in college football history to find the out how many of those wins have been run up against East York Veterinary School. Michigan's docket:

Michigan (19). Total wins: 900. Junk wins: 57 (at least). Percentage of all-time wins that are junk: 6.3. Persistent pigeon: Case Institute of Technology (Ohio), 26-0-1 record. Scheduled junk win that turned into a bad loss: Cleveland AA 8, Michigan 4. What, you were expecting Appalachian State? (Since the game was played in 1891, the AA is not believed to be Alcoholics Anonymous.)

That's actually fewer than anyone in the top ten save Georgia, though Georgia did lose to an empty patch of grass in 1943. The worst offender is USC, with 102 Doris Beautician School beatdowns.

Meanwhile… 8-4, huh? Sounds preview-licious. If only I had known about this Cleveland AA game before Michigan played State.

Is that an axe between your legs or are you just happy to see me? Oh, it's actually an axe.

Via Heiko and his insatiable appetite for panoramic photos. Larger version.

Get there. All weeks are championship weeks:

"Probably just as much as you guys are thinking in your heads," Mealer said. "As much as you guys have been hearing it, we hear it even more. And for you guys, it's probably overbearing and redundant and things like that -- I'm sure, because I can see all the smiles when I brought it up -- but for us, that's why you come to Michigan.

"For us players, it's not redundant, it's not, 'OK we get it.' It's, 'Remember last year? We lost a game here, we lost a game there, and we kept ourselves out of that game.' So we accept the constant reminder of it."

Mealer said there's a picture of the Big Ten championship trophy in the team's Schembechler Hall meeting room. There are roses painted on the walls, a not-so-subtle reminder of the bowl that awaits the league's champion.

This one maybe a little more than most.

I'm not sure I can do this any longer, @umichcompliance. I started following Michigan's compliance feed for some reason. Maybe I needed more dry statements of things I can't do and have never even thought of doing in my life.

It's been a passionate affair so far, but I'm not sure if I can deal with the mindboggle on the regular. To wit:

Michigan Compliance ‏@umichcompliance

The NCAA prohibits the sale of any item with the name, picture or autograph of a currently enrolled student-athlete.

That seems like a pretty reasonable restrict—

!!!

Let's not forget, Dude—that keeping pictures… uh… items with the name, picture, or autograph of a currently enrolled student-athlete… on, uh, your official, you know, website… that ain't legal.

Burkhead may not play. Sounds like Nebraska tailback Rex Burkhead may sit out Saturday after tweaking an ACL sprain last weekend:

Huskers coach Bo Pelini said this aggravation isn't as bad as the one in the Ohio State game, but the repeated issues have him inclined to rest Burkhead for a week. Pelini said Burkhead is day to day and has improved since Saturday.

"He felt really good going into the [Northwestern] game, didn't have any issues in practice," Pelini said. "He's frustrated and obviously disappointed he's had to deal with it. It's been too bad."

Backup/co-starter Ameer Abdullah hasn't ben much of a drop-off: he's averaging almost 6 yards a carry on 86 attempts, with impressive output in the three Big Ten games he's played in—5.5 YPC. He's faster than Burkhead, albeit slighter. Former top 100 recruit Braylon Heard should also get some carries.

Also in Nebraska, reviews of their latest game.

Pitch perfect. The MZone is actually written by Stephen Colbert.

"Nation, I've always been a big fan of Michigan State football coach and scowling man most likely to tell kids to get the hell off his lawn, Mark Dantonio.  His enduring sportsmanship, like getting into public feuds with 21 year old college kids by mocking their height, along with always looking out for the health and best interests of his players, evidenced by holding them out at least two plays when they get an on-field concussion, have made him a role model Spartan fans can be proud of before, during and after the burning of their couches."

Fourth line FTW. The top takeaway from Friday's one-off hockey game against Bentley was the fantastic play of what was nominally the fourth line of Andrew Copp, Zach Hyman, and Justin Selman. They scored twice, dominated play when on the ice, and must have locked themselves into additional playing time as Michigan kicks off their final CCHA campaign against Miami:

The line of Hyman, Copp and Selman has “really given our team a life,” according to Michigan coach Red Berenson.

“He told us what we needed to do to have a good week in practice,” Selman said. “He was making sure every day we were going as hard as we could. We kind of worked off what he was doing and followed in his footsteps.

Copp came from the same lineage as Danny Fardig—the guy on the USA NTDP roster who bounces from the U17s to U18s to fill in on the fourth line and never score—but has shown some soft hands early. Hyman played with a high energy level, as well. I haven't seen Selman enough to really get a read on him yet, but the line's performance speaks for itself.

Clare will return to the lineup against Miami, BTW. Mike Szuma was okay as a sixth defenseman on Friday. Michigan Hockey Net has a Bentley recap.

Siberia also FTW. Preds winger Sergei Kostitsyn is spending the lockout playing in Siberia. This is what Siberia looks like. In May.

Kostitsyn, 25, who joined the Predators from the Montreal Canadiens in June 2010, also said he hated life in North America and labelled Columbus "the gloomiest" city in the States.

Unfortunately, he could not continue ripping on Columbus because Omsk was invaded by thousands of red plastic figures at that very moment.

Thinking alike. BWS picture-paged the zero-yard Denard run that featured as an exemplar of Michigan State going ape on run plays in the just-posted edition on this site. Yeesh:

I think the only deep ball in this game was out of an I-Form to a double-covered Jeremy Jackson. Gotta loosen those guys up—yeah, you might turn the ball over if you bomb it deep but if you're just punting in three plays anyway…

How Purdoed it. Doesn't work as well as "Purdon't," I guess, and is equally dumb. Anyway, Ross Fulton's latest breakdown of the Ohio State offense is even more interesting than normal since it deals with topics often found here in the aftermath of OSU's extensive struggles against the Purdue defense. Purdue loaded up to stop Miller, won a lot of battles on the line of scrimmage, and OSU could not punish Purdue cheating off the slot receivers. This sounds awfully familiar:

The Buckeye passing game was perhaps most successful when OSU could get their hurry-up offense going and catch Purdue in uncertainty. This led to the second Buckeye touchdown. But all too often Ohio State missed available opportunities. For instance, Devin Smith several times had man coverage beat. If Braxton simply leads him to the corner and lets him run under it, it was a potential touchdown.

 

More acutely, Ohio State was not able to punish Purdue for cheating off their slot receivers. It was not simply for a lack of trying. Miller sailed one bubble screen over Corey Brown's head. On the clip above, he held the ball a second too long, allowing the Purdue corner to make a nice play. But other times it appeared clear opportunities were available that OSU did not take advantage of. Traditionally, this has been an automatic check with Meyer and Herman. The offense will continue running the pre-called play but the QB will simply pull and throw the football. Perhaps Meyer & Co have not yet given Miller the ability to do so. But by allowing alley players to cheat into the box, a spread offense quickly loses the numbers advantage it gains from the QB run threat.

That's how you end up with 100 yards at halftime against Purdue. OSU started bashing straight ahead A-gap power from the pistol and running their veer a gap inside the crashing LBs, but Miller's exit cut their response short.

Hey hey!

OSU must find more consistent ways to punish a defense that cheats alley defenders. Whether it is with wide receiver screens or bootlegs, OSU cannot allow a defense that much free rein. OSU must not also let opposing defenses dissuade them from sticking to what they do well. Though this is somewhat counter to the previous point, I believe they work together. The Buckeyes must continue to work to establish their base run game and then use play-action to exploit an opposing defense.

This mindmeld thing is creeping me out.

Etc.: Hero police robot back on duty after 'unstable man' blasts it with shotgun. You will be surprised to know this happened in Nebraska, not Detroit, home of police robots. [HT: Corn Nation.] RAWK HELICOPTER FLYOVER. Mmmmm… Dwayne Bacon. Hundred Level video released. O'Bannon lawsuit update. [HT: Get the Picture.] Getting our mean on at MNB. Nebraska equals points. OSU adds an Oregon home and home to their far future.

Shouldn't Spock be in this?

Hi. Me. Back. So before that much-appreciated vacation, I used this space to talk about constraint theory of offense and provide a rock-paper-scissors matrix for offensive/defensive play calling in various offenses. Today I'm on to Part III, the one where I tell you that Rock-Paper-Scissors is only a fraction of the football head game, because the actual decision trees are far too complicated for even a coach to play all of the interactions, let alone teach them. Moreover, unlike in RPS, or super-advanced-nuclear-capable-canid RPS, there are levels to things: scissors cuts paper better than it cuts (but still cuts) woven kevlar.

Rock-Paper-Scissors is a game you learn to play on the bus ride to school in 2nd grade or thereabouts. It is a very simple, 2-dimensional, triangular matrix:

…meaning every point interacts with every other. It's one dimension past a coin flip but you still only need to remember three interactions (yellow lines). Based on your personal capacity for testing the limits of social institutions, you either very quickly or eventually tried to insert an additional dimension to the triangular matrix, and realized that you were exponentially increasing the amount of interactions you had to remember.

Your 2nd grade mind didn't draw this; it just exploded the same way it might if you interlaced Grbac to Howard, Wangler to Carter and Robinson to Roundtree into the same video. Then it came up with a brilliant way to add a point without adding dimensions:

Bazooka!!! Bazooka blows up rock. Bazooka turns scissors into mangled slag. Bazooka leaves only scant streaks of carbon where once was paper. This idea is not totally silly, since defensive coordinators call bazooka like all. the. time.

Bazooka = Vanilla

Just as the offense wants to get really good at one basic thing and then do that thing all the time, many defenses are deploying the same concept. It's a bit harder for them because they have to react to various offenses on the schedule and various plays, but the concept's the same: the defense wants to stay in a scheme that is basically sound, and will mix in blitzes and different coverages as constraints, so that they can keep running their well-practiced, mostly sound blanket defense. Bazooka is a jack of all trades, solid against the run, solid against the pass, solid against tomfoolery, and vulnerable only to great offensive play and their own physical/mental limitations.

Defenses are a bit more varied than offenses but the most popular vanilla D these days, as I mentioned in the earlier discussions, is a Cover-2 zone (above-left) against run-first teams, or the Tampa 2 against pass-ier teams (above-right).

The difference between those two is in the MLB's coverage duties—in a Tampa he has the deep middle, in a regular Cover 2 he has the short middle and can be more involved in the run game. Everybody, including the cornerbacks, are hovering around close enough to gang-attack running plays like a pincer; and soft spots in the zone (everyone has them) are relatively small and difficult for non-accurate quarterbacks.

If these guys are all reacting correctly and aggressively to the run, if the cover guys are fast enough to close their zones, and the four rushers can generate pressure with regularity, this defense can bazooka anybody's rock, paper, scissors, candle, Vulcan, or whatever. Of course that is way easier said than done—remember offenses are recruiting, training, practicing, and designed for attacking this scheme.

Offensive rock is made to beat defensive bazooka. I want you to look at the above and imagine various offenses succeeding against them. ISO running forces the linebackers to read run, read the hole, and get there in time to take out a lead blocker and lead runner who by design are getting there ASAP. West Coast passing lives in those soft spots under and between the coverage. Air Raids attempt to warp the zones into providing bigger holes by flooding and stretching them. Option running makes a balanced coverage into an effective numbers advantage for the offense at the point of attack. Vanilla defense is made to stop whatever's thrown their way, and offenses gain success by making Mr. Jack-of-All die a consistent bleeding death.

To see vanilla defense in action throw a dart at any recording of Iowa in the last 12 years; their M.O. is to stick to this maniacally. The converse in-conference would be Michigan State. On the way home from Europe this weekend I randomly sat next to MSU LB Chris Norman. Other than "Wisconsin's offense was way tougher than any of the SEC teams," and "lol Michigan's uniforms last year," Norman happily admitted "YAARRGGH SPARTY SMASH!!" is the coaches' favorite blitz, and that they'll run it or something like it more than any other play. Some teams like paper.

So there are exceptions but the exceptions can be beat with your properly executed scissors. The point remains that all matrices aside, much about football comes down to defeating your opponent's bazooka, or vanilla thing, or "rock" defense with your rock. If you recognize this particular bit of wisdom from DeBordian philosophy, well yes in this DeBord is absolutely right. But if you don't properly mix in your constraints, and you always run rock to the same spot/guy out of the same formation, and you shuffle your fullba…uh let's not go there.

Next time (last one? I think it is but I keep stretching these): What's Our Rock?

This is a follow up to to Doctor Rocklove a few weeks ago, where I identified the influence of offensive sets on philosophies. If you're not familiar with offensive theory you should go back and read that. If you're a football coach you are welcome to pinch the bridge of your nose and shake your head, for this is only going to cover about 20 percent of what you know to be the basics of offensive football.

The point today is to look at some of the base plays of various offenses, and a few of the constraint plays that they use to counter, and what defenses do to counter that. In doing so I hope to find stumble upon a better explanation of Borgesian offensive theory than the "grab bag" this space has previously suggested.

That Thing You Do

You've probably read enough college football boilerplate by now to have heard a coach talk about "what we'd like to do." This does not have to mean one play, but it often means one concept—very much like a play—which the team will be able to execute to perfection against the defense they want to see. That play is usually going to be low-risk, and if executed flawlessly against the vanilla defense it's built to beat, it will gain a consistent 5 to 7 yards. It can be run out of many formations, and you will practice it a thousand million times until you are sure it will work every time unless the opposition "cheats" to beat it.

For Vince Lombardi it was the sweep. For Wisconsin (and virtually every high school in our division in the late-'90s) it was the ISO. It could be the Triple-Option (Bo), or the Zone Read (Rodriguez), or Hitch-n-Out (Walsh) or Levels (Peyton Manning's favorite), or 62 mesh (Captain Leachbeard). With passing offenses, which is Borges's thing, it's important to note that the core concept itself can often be a package of plays which work off of each other, none particularly favored; for running the same concept will vary where on the line it will attack.

You can go crazy for your core concept. You can practice it incessantly. You can recruit players whose skills best fit what they're supposed to do on that play. You can even focus physical training on developing muscles that are used on that play. The better you are at that play—and this is a sliding scale—the more the defense has to move someone or do something to "adjust" to you. But this is a zero-sum game, so if you're moving a defender to stop the base play, he's no longer doing the thing he was doing before. He is making something else way easier than it should be. He'll do this anyway, until you make him pay.

Constraint Theory of Offense and RPS

What you choose as your core play or concept will determine much about the other things your offense does, because now you add plays to punish defenses for adjusting to your base play. That's what coaches mean by "constraint"—you are constraining the level to which the defense can react to your bread and  butter. What you are essentially doing is creating an environment in which you get to run your core play, which you've practiced more than any other play, exactly how you drew it up as much as you can.

Certain concepts are almost always constraints because they won't work against vanilla defenses. Delayed handoffs work because the defensive line is closing on the quarterback as if it's a pass play. Halfback screens work well against blitzes but if a linebacker is in man on the running back, a vanilla defensive concept, you're screwed.

Defensive wins in rock, paper, scissors are rare and lucky guesses; usually a D's successes come from outstanding execution of a vanilla defense, for example if the nose tackle shoots past a playside block and forces a pitch on a speed option (as if that could happen).

Defenses have constraints too but theirs are limited by the offense's greatest advantage: whoever has the ball chooses the play (the D's advantage is so much more can go wrong with offensive execution). Defensive constraints translated to boilerplate sound like "we took away the run and made Denard beat us through the air." What they mean is the defense was cheating against the offense's base play all game but leaving themselves more open to the constraint plays, betting on poorer execution by the offense.

Dantonio last year sent two blitzing linebackers up the middle on many occasions, taking away Michigan's bread 'n butter play "Denard-'n-stuff." This forced Michigan into our constraint, which was targeting open receivers in short zones, but then Dantonio took this away by having safeties replace the blitzing LBs. This opened up another constraint by making deep coverage completely up to the cornerbacks, but then a trash tornado covered that constraint for them.

What the constraint theory does for playcalling is create a kind of matrix of offensive adjustments to defensive adjustments and adjustments to those adjustments. For a typical varsity high school team that matrix is probably 20 plays, and for college football it's more complex, and in the NFL the adjustments are so myriad and subtle I'd have an easier time teaching EMI/RMI shielding (it sounds hard).

Because the shades of gray in such a big decision matrix make for convoluted understanding, I've tried to (over-)condense the basic constraints of four basic offenses. There is way, way more but these are a few of the constraint packages that Michigan used last year.

Manball

Offensive Concept: I'm bigger, faster and stronger than you are, so I'm gonna hit you so hard your momma cries, then evoke masculine metaphors.

Defensive Concept: Control the point of contact, win 1st down, never let the train leave the station.

Timed Passing (West Coast)

Offensive Concept: A symphony of route design and timing that puts defenses into a progression of impossible choices.

Defensive Concept: Throw off your timing, suffocate your routes, kill your conductor.

Read Passing (Air Raid, Pro)

Offensive Concept: Spread, mesh, read, and gun, so on any given play, at any spot on the field, we can put it where you ain't.

Defensive Concept: Anywhere you can get, I can get faster

Option (Triple-Option, Zone Read)

Offensive Concept: Reverse the traditional 7-on-6 "numbers" advantage of the defense in the running game (i.e. their front 7 versus 5 OL and a running back) by having the quarterback participate, and "blocking" an edge defender by optioning off of him instead of wasting a body.

Defensive Concept: The cat has more patience than the mouse.

Next time in this series: vanilla defenses, and the best offense for Michigan this year and beyond.

Last time on Picture Pages we looked at a 35-yard iso on Michigan's first drive against Minnesota. A 35 yard iso means someone did something terrible on defense or your tailback did something ridiculous; Michigan was playing Minnesota so it was the former.

So Michigan scores a touchdown and gets the ball back and opens up with the same set. Minnesota again has both safeties rolled up.

On the snap the line pass blocks but the backfield executes a historical reenactment of The Battle Of Minnesota Sucks At Isos.

There's a gap in the line that Hopkins is thundering towards again and by the time it's clear Denard has the ball the three LB type substances have started moving towards the LOS:

When Hopkins hits the LOS the two guys who could hypothetically cover him are four yards from the LOS and stationary.

This is what it sounds like when doves cry.

Video

Items of Interest

Constraint theory right here. Minnesota just got hit with a big iso and got chewed out on the sideline about it. They are hyped up to stop it, so when Michigan shows it again the LB and S suck way up and leave Hopkins open for a big gain over the top. This works not only because Minnesota overreacts to it but because of the omnipresent Denard threat posed in the shotgun. That means the Gophers are operating with essentially zero deep safeties.

This is what happens when you can force the defense to cheat. An actual opponent playing this way probably would have given up five, not 35, on the first iso, but that's enough to force them to cheat to it, whereupon bang.

This isn't unique or new. Literally every program in the country except Ohio State* tries to do this. Michigan's old-school waggle is an example. The hope with Borges is that he'll use them frequently to score lots of points instead of occasionally as part of a bler offense run by 70s thinking. Denard is a hell of a thing to try to stop without cheating, much more threatening than Michigan's four-yards-and-a-cloud-of-dust running game was in the late 90s and early aughts.

This is the stuff you get in the shotgun even when Denard is not running because the zone read demands attention at all times. Under center Denard's legs mean waggle or Incredibly Surprising QB Draw, neither of which forces safeties into the box.

We saw a bucket of constraint against the Gophers. This is in part because the Gophers are so bad they have to cheat every play in the hopes of stopping the opposition.

I think it's also in part because Borges is finding his legs in this strange environment where your quarterback is one of the most threatening rushers in the country. Michigan tried the waggle a bunch earlier this year (and in spring) and got little out of it; against the Gophers that was ditched in favor of plays that use fear of Denard's legs as a major component of their success.

Hey man, nice touch. Denard floated a nice catchable ball to Hopkins about 18 yards downfield. He could do that because there were no safeties, so I've got no problem with the throw.

Let's see if Shaw can block this. Because this is a 75 yard touchdown if a fullback isn't running it. Also Hopkins almost dropped this.

*[

]

One of my early complaints about the Denard-Borges fusion cuisine was the grab-bag nature of the offense. By that I mean the sense that Michigan's plays were generally unrelated to each other and worked because they were new or the opponent was poor, not because they put the defense in a bind trying to defend one thing while another was happening. You can only run throwback screens out of an ace set a limited number of times when you don't roll the pocket out of an ace set effectively; you can only run a quick pitch that plays off a FB dive a limited number of times when you never run the dive.

That complaint is increasingly invalid as Michigan refines what it does. Full Minnesota disclaimers apply, but the most encouraging thing about last week's game other than everything was the series of gotcha plays that gashed Minnesota. BWS did a great job of showing how Michigan's long-overdue deployment of the sprint draw* (in this case a bonafide counter with a pulling LT) looks just like the QB run game that has been the heart of Michigan's offense for a year and a half. The sprint draw is a constraint play that punishes you for cheating on the offense's bread and butter.

That's one example. The Fritz package is another example. Michigan got a speed option blown up the first time; when they came back to it they ran a quick pitch that played off that option. This is what it looked like:

Check that safety on the far left hauling ass to the presumed option side. He gone. By the time Toussaint hits the corner ain't nobody here but us chickens:

Minnesota is exceptionally bad at all things but this is the kind of stuff that gives defensive coordinators hives. That looks just like OH CRAP DENARD OPTION until it's too late.

But wait, there's more! If you were surprised when Michigan opened up its second drive with a lovely touch pass from Denard to Stephen Hopkins, that makes twelve of you. He'd set Minnesota up for it on the previous drive.

*[I do have a slight disagreement w/ that post, FWIW: On that play it's clear Huyge is expecting to kick out the DE. When that DE comes inside rapidly Huyge looks like he's losing him. Lewan is supposed to hit the backside B gap, which has a marginally blocked guy in it. If Lewan doesn't block the DE there's a chance he shoots up into Shaw for a loss. I think you leave the safety for the RB.]

Play The First: New School Iso

It's first and ten on the Michigan 38 on the first drive of the day. Michigan comes out with what is for them a power set: shotgun with two backs and a tight end. Minnesota rolls both safeties to 7-8 yards and plays way off the WRs.

They're going to run an iso off the right side of the line. Iso kind of looks like inside zone—no one pulls, you try to combo defensive linemen—but you get a lead back roaring up in a designated hole. On an inside zone a blocking back will usually flare out or head backside to provide another gap on one end of the line and the running back will read his blocking and pick a hole.

Here it's straight upfield, hole or no. This train is headed A-gap.

It's Minnesota so there is a hole. Schofield and Molk send the NT to his knees. Omameh locks out the other DT and Denard holds the backside end with the threat of his run. A crease forms in the intended spot:

Hopkins thunders into it and lowers the boom.

And that's all she wrote. The two DTs getting annihilated and Hopkins thumping the MLB such that he provides a crease away from the Gopher free hitter—visible in the left frame above and stuck behind the Hopkins block in the second—gives Toussaint a free pass into the virtually nonexistent secondary.

Note that Molk is still waiting for someone to block. Minnesota is not good.

Toussaint runs through a diving tackle attempt and is eventually run down because he has to break his stride to do so. 35 yards.

Video

Items of Interest

Minnesota is awful. I award them no points, God have mercy on their souls, etc. Not much else to say.

On this play three separate Minnesota defenders are crushed by their Michigan counterparts and Molk is just like hanging out because the Gopher LB is hanging around on Robinson when Robinson is being contained by a DE. Against a real team this is an eh gain.

This works for a lot of reasons but the paramount one is the Hopkins block. This is awful Minnesota play, but Hopkins makes it count by getting a driving block on the LB that kicks him out of the lane. If the guy gets inside of Hopkins Toussaint cuts out into an unblocked safety and picks up five or so yards unless he makes him miss; even if he manages that the process of making him miss will probably get him tackled by the backside DE.

But Hopkins lowers his shoulders and lifts the LB out of the hole, eliminating two guys and turning this into a big gainer. Without one guy eliminating two you can't pick up a bunch of yards when an extra safety is in the box*, especially on an old-timey quien es mas macho play like an iso.

*[And by "an extra safety" I mean two extra safeties; Denard + shotgun == extra guy in box is standard. Here both safeties are rolled into the box.]

Don't get down about Toussaint's speed because of this play. Yes, tackled from behind by a Gopher, but the ankle tackle he ran through put him off balance and slowed him up; without it this is likely a touchdown.