Introducing “Multi Shells” (Bob Overing)
The Problem
Last season I complained about the rise of what I (and maybe others) call “combo shells” or “conjunctive shells.”[1] I spent several hours and phonecalls with some of my favorite LD debate theorists like Adam Bistagne and John Scoggin thinking about how to treat combo shells. I revisited the metatheory argument that Jake Steirn, Martin Sigalow and I worked on to deal with these shells at the 2015 TOC. I still don’t have the answer, but I have a new idea!
To recap the puzzle: some combo shells seem legitimate, such as “no conditional PICs,”[2] but it’s very unlikely that theory arguments like “neg can’t read the Hobbes NC, T, deny the aff the RVI, read the NIB that they read, and turn the aff” isolate unique, structural abuse beyond their component parts. A very common trick for the 2AR is to say something like “sure, you’ve proven that each position alone is fair, but my strategy skew argument states that the conjunction is unfair, so proving the fairness of each part is non-responsive.
But of course, proving the fairness of each part is responsive. All else equal, a five-pronged neg strategy where each prong is itself fair seems more likely to be fair than a five-pronged neg strategy where the fairness of each prong alone is in question. Taking the 2AR’s line seriously, that only the conjunction is unfair, you have to believe that each prong is a totally permissible debate strategy. And any two of the prongs together would be fine too. Three of the five? Go for it! Heck, if you had done just the first four without the fifth, we’d have no problem. But all five? Now that’s cheating!
What I’m getting at is that I don’t believe debaters and judges are evaluating these shells properly because they aren’t seriously supposing that each prong of the strategy is itself good and fair. Or at the very least, they aren’t weighing the (un)fairness of each prong in assessing the fairness of the conjunction. If you find the strat skew argument compelling in the example of the above, it’s likely because you really think NIBs are unfair or RVIs on T are good. That is, lurking in the background of judgments about combo shells are judgments that some of the shell gets it right. If this is true, a judge isn’t really voting on the combo shell, but a hidden NIBs bad or RVIs good argument (s)he does find compelling.[3]
Given modern theory practices in LD, this is wildly inappropriate. For one, it hides the ball so the 2NR doesn’t know what the theory debate is really about until it’s too late. Two, the structure of these shells does not currently accommodate this type of analysis, since a judge cannot vote on NIBs bad if the interpretation is not simply about NIBs. The “trick” cuts both ways since NIBs could be unfair but the conjunction still fair.
The Solution
My solution is to make explicit the intuitions lurking in the background of combo shells by forwarding each part of the conjunction as its own theory argument. In the status quo combo shell model, our example would be fleshed out like this:
A, Interpretation: Negatives may not read a Hobbes NC, T, no aff RVIs, a NIB, and turns to the aff.
B, Violation: They did.
C, Standard: Strategy skew. The 1AR is so hard if I have to answer T and don’t win for answering it, beat a NIB and don’t win for answering it, prove my framework is better, and answer case turns. It’s impossible in four minutes! The conjunction is particularly unfair because… blah blah blah [insert more whining]
Under my new model, our example would look like this:
A, Interpretation: Negatives may not read a Hobbes NC, T, no aff RVIs, a NIB, or turns to the aff. Nor can the negative read the conjunction of the five.
B, Violation: They did.
C, Standards:
(1) Hobbes is unfair… [e.g. for ground or any typical theory standard]
(2) T is unfair…
(3) No aff RVIs is unfair…
(4) NIBs are unfair…
(5) Turning the aff is unfair…
(6) The conjunction is unfair… [insert the strategy skew reasoning above]
Call this a “multi shell.”[4] The example may seem silly because obviously Hobbes, T, no aff RVIs, and turning the aff are fair, but that’s the point. If four of the five parts are fair, it should be acknowledged that the strength of this shell turns on the success of standards (4) and (6). If there is no argument for (1), (2), (3), or (5), the shell is very weak, and this structure makes that apparent. So a bad combo theory initiator wouldn’t want to utilize this structure; (s)he would want to obfuscate the issues at stake.
But there are times when one would utilize this structure. You might think each component of your shell is very strong. To use our earlier example,
A, Interpretation: Negatives may not read a conditional counterplan or a plan-inclusive counterplan (a PIC). Nor can the negative read a conditional PIC.
B, Violation: They did.
C, Standards:
(1) Conditionality is unfair…
(2) PICs are unfair…
(3) Conditional PICs are especially unfair…
Each of (1), (2), and (3) are plausible theory arguments, so why not separate them? To my knowledge, no one is forwarding theory arguments like this, but it has clear strategic advantages in granting the multi shell initiator three ways to win in the 2AR; each theory argument alone can be a voting issue on my model. Further, the technical objection to the structure of combo shells is eliminated by the disjunctive phrasing; the interpretation is three interpretations in one, and the 2NR’s defeating one of them does not necessarily defeat all three. This multi shell model is also superior to initiating three separate theory interpretations, since there is time lost in transitioning between flows and repeating the shell structure each time. Finally, the model could be utilized by theory respondents to clarify the debate, e.g.:
A, Counter-Interpretation: Negatives may read a conditional counterplan, a plan-inclusive counterplan, or a conditional plan-inclusive counterplan.
B, I meet.
C, Standards:
(1) Conditionality is fair…
(2) PICs are fair…
(3) Conditional PICs are fair…
I believe this “multi shell” approach is both strategic for theory initiators and respondents and provides much-needed clarity to combo shell debates. And even if the structure is not used, judges should have each of the components in mind when deciding these often complicated debates.
Questions for Further Research
- Should we give any weight to responses to combo shells that take this form: “PICs are so often read conditionally that all of my PICs good arguments are disadvantages to the conjunctive interpretation, since it would eliminate much of the instances of PICs.” Does whether we like this response depend on whether we view theory as “norm-setting?”
- Try to imagine a criminal justice analogue to the combo shell, i.e., a conjunction of acts where each is legal, but the conjunction is or ought to be criminalized.
- Last year I wrote several “Holiday Disclosure Posts”; what should I spotlight during this year’s break?
End Notes
[1] I think some people call them “multiplank,” but I’ve never quite understood what a plank is, so I’ll stick with combo/conjunctive for now. Plus, I’m going to re-appropriate “multi”
[2] The reasoning would be something like “conditional PICs force the 1AR to debate on two fronts. Against a conditional counterplan, the aff must leverage its advantages against the CP and the status quo, which are very similar tasks, but against a conditional PIC, the aff must leverage a tiny part of the aff (the part not solved by the PIC) against the net benefit and also the whole aff against the status quo, two very different tasks. This makes high-quality analysis unlikely given 1AR time constraints, and the 2NR has all the power to decide which of the two very different debates is decisive, rendering moot the best 2AR option.”
[3] Another possibility is that judges have some sympathy for a very tough 1AR, and the strategy skew claim encapsulates how hard it is to be aff against a good neg debater in contemporary LD. This line is inappropriate too – yes, affirming is harder, but that’s not a reason to penalize the neg for pursuing a killer strategy.
[4] Open to suggestions.
4 Comments
Two things strike me about this proposal.
First: The advantage to reading a “multi shell”[1] (viz., it’s three potential theory outs) might also be a disadvantage, since each separate abuse story requires a substantial time investment. Rarely do theory debates involve only one standard per shell[2]. Developing three (at a minimum) separate abuse stories is a tall order, especially for shells where one story is much weaker than the rest. In the case of conditional PICs, for instance, it may be truer (or at least more strategic) to only argue that the negative can’t have access to conditional CPs, and that PICs are a worse instance of a conditional CP meriting separate treatment[3]. There seems to be no reason, other than a vague notion of “clarity,” to require that theory initiators construct the most restrictive disjunctive interpretation possible.
I have three complaints with this “clarity” argument, besides the vagueness complaint. First, it’s very confusing trying to weigh one quite separate abuse story against another. How does the badness of reading a conditional CP compare to the permissibility of reading a PIC? They impact to the same standards, but in such a causally and logically independent way that it’s hard to sort out which link is stronger[4]. You might object that, on your model, the theory initiator can go for just one of the parts of the interpretation in later speeches. But that just pushes the comparison one level higher. What if the 2NR wins an RVI on PICs good, but loses the conditionality debate? Second, the ability to kick shells with impunity has massive strategic value. Your model advantages theory initiators, since respondents have to hedge against more possible outs. Finally, aesthetic clarity in theory debates is probably less important than strategic considerations: judges should not generally impose strategy upon debaters, and a 1AR might easily decide that it’s worth it to sacrifice the twenty seconds it may take to prettily explain one-third of an abuse story in order to cover case. The 2NR can always bring up that third, if it’s relevant.
Your best response is to reiterate that the truth of the combo shell depends on the truth of each component (e.g., that we should not presume it’s abusive to read a conditional PIC if the 1AR only warrants why conditionality is bad, and tacks on a claim about PICs). But this brings me to my second major thought: it is not obvious that “condo PICs bad” assumes all of
(1) “PICs bad” and
(2) “condo bad” and
(3) “condo bad in the case of PICs.”
First, (1) and (2) already entail “condo PICs bad.” If PICs are bad (in all instances), and if conditionality is bad (in all instances), then surely a conditional PIC is also bad. You may respond that debaters often do not warrant that “in all instances” addendum (indeed, many counter-interps are of the form: I may violate the interp under a certain condition), but that is not relevant to the semantics of the interpretation. Moreover, there are many possible ways to get to “condo PICs bad.” For instance:
(1a) PICs are bad in at least some instances and
(2a) Conditionality is bad in exactly those instances where PICs are bad
suffices. So does (1a) combined with:
(2b) Conditionality is bad in at least those instances where PICs are bad.
Obviously, not every combination of scope quantifier and predicate works. For example, combining (1a) with
(2c) Conditionality is bad in at least some of those instances where PICs are bad
or with
(2d) Conditionality is bad in at least some instances
does not yield “condo PICs bad.” Additionally, (3) seems already to yield “condo PICs bad.” There are, as you acknowledge, at least some cases of genuinely unique abuse stemming from the conjunction of two or more practices[5]. Finally, reading a conditional PIC might simply be a separate practice in its own right, with a unique set of theory concerns (viz., exactly those concerns that arise from reading a PIC conditionally, and only some — or none — of the concerns that arise from merely reading a PIC or reading a conditional CP). Debate specifies the structure of some actions (like a reverse voting issue, or kicking a CP, or triggering a contingent standard), in the same way baseball rules make intelligible certain actions, like stealing a base. Who’s to say, apart from the norms and rules which govern debate, what exactly those specified actions are (and are not)?
Notes:
1. I always called these “story time shells,” since often the strategy skew standard would be a long tale about just how hard the speech is, and just how sympathetic the judge should be to the poor aff debater. But FWIW my vote’s with combo shells.
2. Moreover, abusive strategies are often abusive for multiple reasons, at least at the level of analysis at which many theory standards operate. For instance, NIBs bad, probably one of the better candidates for a cut-and-dried reciprocity argument, is commonly read with an additional ground or a strategy skew standard. Perhaps NIBs on this topic flow one way or the other, or negative NIBs are more preclusive generally than aff NIBs (since the negative gets arguments like skep and various forms of nihilism, while the only similarly total aff argument is trivialism, which may not clearly affirm at all). NIBs might also unfairly alter the respondent’s strategic calculus; since they are so one-sided and preclusive, respondents might have to invest a substantial portion of their speech — much longer than it took to develop the NIB, proportionally — to counter the strategy. T is another good example. Affs with a tiny literature base might have bad negative ground; they also are likely to be unpredictable and to generate less ‘core’ topical education. I make use of a notion of “level of analysis” because I think it’s sometimes helpful to analyse abuse stories slightly differently; part of the reason ground/predictability abuse on T is often articulated as a limits standard is both the ground claim and the predictability claim stem from the same source, viz., limited literature on the plan.
3. You might object that the reasons a conditional PIC is worse than a conditional CP are just reasons PICs are bad in disguise. I’m not sure that’s necessarily the case, but even so, the force of a reason can vary by circumstance. PICs in general might be good (say, to achieve a nuanced discussion of certain specific areas of the topic literature) but those reasons might just not *apply* if the PIC’s conditional (that discussion is unlikely to occur if the neg can just kick the CP in the 2NR). Or, an abuse story can become stronger in a certain context. For instance, one of the problems with affirmative framework choice (AFC) is that the aff can pick a framework under which there is little neg offense, pigeonholing the neg into defending a weaker argument. Against a util aff that objection might be unconvincing, but against an international law aff it is much more compelling.
4. This is, in fact, one of the reasons theory interpretations exist in the first place: to clarify the terms of the theory debate. Otherwise, we might simply say something like “Interp — Neg must be fair” and make each shell a separate violation. I’m not convinced this is entirely a bad idea (might take this up in a future post), but it’s at least a stronger case of my confusion objection.
5. To answer your law analogy: you may ordinarily be able to spray permanent aerosol paint on walls (say, those walls that you own), and you may be able to spray some substances on walls that you do not own (say, compressed air), but law typically proscribes graffiti. I think this is not a very helpful analogy, though, for a couple reasons. First, laws against graffiti are codified by an authority (the criminal code of a state), while theory interpretations are not. Second, law can give rise to a certain description of action — say, of vandalizing a store window. Outside of law, we might have an entirely different description of an act of vandalism: it could be part of a clamorous protest on Inauguration Day, or it could just be a special case of smashing a window. Debate operates similarly, what with its jargon and technical focus, but combo shells could plausibly describe a NEW practice, as below.
*re time investment*
Yes, there’s a opportunity cost in the 1AR to advancing multiple theory arguments. But this model does decrease that cost relative to reading multiple shells. And if you’re negative, especially dealing with an insidious combination of spikes, a multi shell might be a good option.
I think the condo PICs example is a time when it pays to read a multi shell. There’s one good fairness argument and one good education argument for condo bad — that’s 12 seconds for an additional voting issue. I think the same about PICs. So why not add 24 seconds for two additional theory outs, assuming one can win the RVI debate (or they’ve conceded a no RVIs spike)?
Your end note about NIBs bad isn’t really an objection to multi shells. I certainly did not mean to imply that multi shells are always strategically superior; NIBs might be a case where it pays to have depth over breadth of theory offense, especially if the violation is strong.
*re clarity*
The fact that it’s very confusing is true in the status quo. If I’m right that PICs good and condo good implicate the truth of condo PICs bad, that evaluation will occur whether the debates are separated out or not — it’s just easier this way. This also answers your worry about comparing a world where the neg and aff have each one parts of the debate. In any event, it’s no more convoluted than introducing three separate theory arguments, which isn’t all that ludicrous in the current meta.
You anticipate my response and detail how a condo PICs bad shell might not logically rely on assumptions about the theoretical legitimacy of condo or PICs. Of course, condo PICs is my example of a good combo shell since there is a story about unique structural abuse due to the combination (as I explain in end note 2). So in that case, it’s not obvious that much clarity is gained by separating out the components.
I can concede that much, but it’s really just “defense” as far as my central thesis is concerned. It may be that there is a unique argument for (3) that does not depend on (1) and (2). I say that (a) often there is not, so it’s valuable to separate (1) and (2) from (3) in those cases, and (b) even if there is a unique argument for (3), it may be strategic to separate.
Step away from the formal logic for a second and think about the substance of these arguments. If the neg says conditionality is good because it presses the aff to defend itself in different ways within the same debate, that argument mitigates the unique abuse story in (3). If the neg says PICs are good because they force the aff to think on its feet in developing a new ballot story, that argument also mitigates the unique abuse story in (3). I am not claiming that it is impossible to construct an argument for (3) that does not depend logically on (1) and (2). I am claiming that in practice, that occurs far less often than debaters and judges admit.
I think the best model is still just reading the combo shell.
**time investment:
Surely not every 1AR can afford to spend an extra 24 seconds. Part of what makes 1AR combo shells strategic is they already incentivize the 2NR to hedge. For instance, a strategic 2NR will argue against “condo PICs bad” by saying:
(1) The neg should *always* get to read a PIC,
(2) The neg should *always* get to defend (one) conditional advocacy, and
(3) Condo PICs are independently good.
It might be a good idea, occasionally, to open up a potential ballot on PICs bad and condo bad. But, “PICs always bad” is still offense to the interp, since the interp allows fewer PICs than the counter-interp. Also, I worry that too many 1AR outs is too many. Affs should focus on offense they could plausibly sit on in the 2AR. You might object that one fairness and one education argument against PICs is sufficient for a 2AR. But then I worry your model makes the 2NR too hard, which overcompensates.
For the negative, the calculus is still the same. You may have to read the same theory arguments against each spike (if only to prevent that spike from being extended against you). It’s still better, I claim, to read those arguments on the spike itself, if only to make it clear that you’re not dropping anything.
**clarity:
It’s easiest on three separate flows. The extent to which your shell decreases time cost is the extent to which it muddles the debate. You’ll say your model is a good middle ground between clarity and time cost, but the incentives argument I made above indicates those debate still get fleshed out enough.
You say (a) often there is not a unique argument to be made. But in that case the conjunction really is not warranted at all. Then, debaters should just read the individual shells. For instance, “can’t read the Hobbes NC” (perhaps it doesn’t have turn ground). I already answered (b) above.
I think we need to distinguish between strategy and truth: for bad combo shells, separation (at least by the judge) is necessary to evaluate the truth of the interp. For good combo shells, it is not. In that case judges shouldn’t use your components analysis if that involves changing the evaluative framework the debaters have set up (or assume) for that shell. One such evaluative framework could be a norms-setting model, as you point out in a QFR.
**time
In the status quo, tons of teams read condo bad and PICs bad in the line-by-line on the counterplan. In those cases, I’m advocating they add the stronger combination interp and move the two shells into the one, which I’m sure you’ll agree is better for clarity. The time lost in that case is pretty minuscule, right?
I do object that one fairness and one education argument for PICs bad is sufficient for a 2AR. I do not understand your reply that it makes the 2NR too hard — that’s what a good 1AR should be doing!
**clarity
Three separate flows is a huge time loss for the initiator, especially if it’s a 1AR, for the obvious repetition reason but also because many arguments will be cross-applied in the 2AR. If we’re talking about cases where PICs bad and condo bad make a difference to the truth of the combo interp, then those cross-apps will occur, so why not put them on the same flow?
Also, the type of clarity loss that advantages my model outweighs the “muddling” you discuss. The current lack of clarity causes poor debating because it hides the ball (i.e., debaters don’t debate PICs bad and condo bad when they are relevant to the truth of the combo interp) and causes judges to make incoherent decisions. I am more worried about that than judges wading through technical line-by-line, which seems inevitable in these types of debates, and separating flows doesn’t really resolve.
I largely agree with your last paragraph. I explicitly state that multi shells are better for evaluating the truth of bad combo shells, and that condo PICs bad may not be an example where you’d want to run a multi shell.
If I’m right about clarity, then the intervention impact is an advantage for me (turn!). I maintain that judges are intervening NOW in accepting the 2AR “trick” and ignoring arguments for condo good / PICs good in the 2NR that implicate the combo interp. See the examples I gave at the end of my first reply.