The article is good, and I also like your banned and restricted list better than WotC's. That said, a few points:
You mention that this is hard to define, but you don't actually give us your definition.
Intentionally. In my view, "interactivity" is a folk taxonomy, like "race," which can't actually be scientifically defined in Magic, or it quickly begins to break down. I wrote several articles about this many years ago for starcitygames. I won't recapitulate them now, but you can find them easily using google.
So, what I'm really capturing when I say "interactivity" as a factor, is the situations, decks and circumstances that most people regard as non-interactive, to such a degree that it merits DCI intervention. I don't have a definition from that - it emanates from the crowd. And I don't think it can be parsed in a scientifically defensible way.
One way to scientifically define interactivity is this: an interactive decision point is one where cheating by looking at your opponent’s hand will improve your probability of winning.
A game with zero interactive decision points is basically a game of solitaire - all your decisions are independent of your opponent’s plans.
Increasing the expected number of interactive decision points, and increasing the associated expected winning probability improvements, would satisfy those wanting more interactivity.
It depends on how you define "correct." My preference for restricting Power Artifact over Consult in Old School '96 is a perfect example of a counter-intuitive narrow tailoring approach.
As I said above
Narrow tailoring assumes that there are multiple ways to solve a problem (which is usually true when it comes to B&R policy), and that the means by which you accomplish that end should be that which has the least collateral harms. Therefore, narrow tailoring will often lead to restrictions that seem facially absurd, in the sense that the restriction target is a less seemingly 'powerful' or salient card.
Demonic Consultation is used by a number of Old School decks in Old School formats with Ice Age. It makes Reanimator more consistent in Old School '95, and MaskNaught more consistent in Old School '96, thereby diversifying both metagames. It boosts mono black decks and a number of other decks that tend to be homogeneous as well.
On the other hand, Demonic Consultation makes Power Artifact combo decks too good (and even better in Old School '96, when Force of Will is available). Restricting Consult instead of Power Artifact collaterally harms Reanimator, MaskNaught and other decks unnecessarily. The more narrowly tailored choice is restricting Power Artifact, which is the best thus-far-proven use of Consult in environments with Necro restricted, and which used to be restricted in Swedish Old School anyway.
It's possible that Consult is so good that other decks can become too good by abusing it. If it were proven true, then the calculus would shift, and I'd prefer to see Consult restricted instead. But I haven't seen that to be the case in Old School '95 of '96 tournaments. And, part of my methodology is evidence-based decision-making rather than theory crafting around the power of a super-efficient black tutor. I invite folks to try to abuse it, and make a deck that is too good. If they do, then I'd restrict Consult instead. But until then, I'll go with the choice that maximizes strategic diversity while minimizing the number of cards restricted.
@dshin I'm not sure hidden information is the right litmus test for interactivity. I would consider a complex board state with many deep lines available to both players "interactive" even in the absence of any hidden information.
I agree that it’s not quite the right litmus test. What’s nice about the test, however, is that it can easily be formalized mathematically. We can assume the players follow Nash Equilibrium strategies (game theoretic optimal), both in the true game state and the hypothetical game state involving extra information - these have associated winning probabilities that in principle can be computed mathematically.
An alternative test that might address your objection is this: an interactive decision point is one where cheating by reading your opponent’s mind will improve your probability of winning. If the players are game theoretic optimal agents, then the two tests are identical. But if you are a human, they differ precisely due to humans’ propensity to mis-process perfect information.
@ribby It is potentially relevant. Someone might say, “interactivity cannot be defined; thus it is not a valid consideration for restriction decisions”. Demonstrating a definition, however academic, legitimizes the concept as a criterion for restriction decisions. The fact that the definition is purely mathematical shows the criterion can be completely objective rather than a form of mob justice.