A-est N possible [data of the day]

Consider the following sentences:

1. He ordered the strongest drink possible.
2. He ordered the strongest possible drink.

---

Observation: the two sentences seem to be semantically equivalent.

Question 1: are the two actually meaningfully equivalent? Are there contexts where one is felicitous and the other isn't? Does one result in an additional implicature/presupposition/piece of meaning that the other doesn't?

---

Observation: 'possible' does not seem to modify drink. Notice that when 'possible' would unambiguously modify drink, as in the latter two of the three possible relative clauses, the meaning is changed:

3. the drink that he ordered was the strongest possible drink.
4. ≠ the strongest drink that he ordered was a possible drink.
5. ≠ the possible drink that he ordered was the strongest drink.

(It is not even clear to me what a possible drink is.)

Question 2: What does 'possible' modify? Is it telling us to consider all possible worlds?

---

Observation: there seems to be a relationship between '-est' and 'possible'. Removing '-est' renders the sentences ungrammatical.

6. * He ordered the stronger drink possible.
7. * He ordered the stronger possible drink.

Question 3: How do we formalize the relationship between '-est' and 'possible'? And how do we derive the two word orders? It seems to be the case that the word order alternation is dependent on '-est'/'most', given data like the two above.

---

Observation: it is not just '-est'; 'most' can also license post-nominal 'possible' (though prenominal 'possible' is far less acceptable to my ear).

8. He ordered the most alcoholic drink possible.
9. *? He ordered the most alcoholic possible drink.

And it is not just superlative adjectives:

10. He tried to get the most money possible (from the ATM).
11. ? He tried to get the most possible money (from the ATM).

The latter of the two sounds less good to my ear, but there are hits in corpora / on Google.

Question 4: How does this affect the formalization between '-est'/'most' and 'possible'? It is not the result of attributive adjective syntax that post-nominal 'possible' is licensed. (Though, I suppose this depends on how you treat 'most' in these examples.)

---

Observation: Some other words can occur post-nominally or pre-nominally, only in a superlative context; e.g. 'available':

12. He ordered the strongest (available) drink (available).
13. He ordered the stronger (*available) drink (*available).

Some other words are also restricted to a superlative context, but must occur post-nominally; e.g. 'ever':

14. He ordered the strongest (??ever) drink (ever).
15. He ordered the stronger (*ever) drink (*ever).

Question 5: What determines which words are sensitive to superlative contexts, and what determines where those items can occur in relation to the noun? I have a sneaking suspicion that it's something like movement and the fixed position of adjectives (a la Cinque).

---

In summary, we're left with a lot of questions, and not a lot of answers. That's what research (and not morning musing) is for.

both easy to share and not easy to share

burger king has a new ad campaign.
     (1)  BK minis are easy to share.  But that doesn't mean they're easy to share.

of course, using at-first-sight-ungrammatical/infelicitous slogans (perhaps to shock you into paying attention?) isn't anything new (campbell's "soup that eats like a meal" comes to mind).  but this seems dangerously close to the kind of ling-101-semantics-lecture level of contradiction that should be impossible for any speaker.

so why isn't it so bad?  it seems to me to be a case of two kinds of implicit arguments. they way i believe we're meant to interpret the slogan as (2), where i've put the interpretation of the implicit argument in square brackets:
     (2)  a. BK minis are easy [one] to share.
           b. But that doesn't mean they're easy [you] to share. 

that is, (2a) is a generic statement about the by-design nature of the product small and sharable -- for anyone.  but (2b) is a specific statement about the actual state of affairs for you, the listener.  the being-advertised-to.

perhaps this distinction between the two implicit arguments in (2a) and (2b) is clearer in a variation of (1), as in (3): 
     (3)   I know BK minis supposed to be easy [one] to share.  But that dosn't mean they were easy [me] to share when I bought them yesterday.

to be clear, the second sentence cannot involve a "I" controlling a PRO (there is no c-command in this case).  it seems that this context just facilitates a generic interpretation in the first sentence, and a specific one in the second.

these implicit arguments would seem to be what are generally put under the umbrella of PROarb.  that is, a PRO-like silent category that can have arbitrary reference.  what i think (1) shows is that PROarb can come in two different flavors: generic or specific.  does that mean that we need two different lexical items, PROarb-gen and PROarb-spec?  or, does semantics/pragmatics distinguish the two in the relevant way so that (1) is not a contradiction?

i don't know much about the PROarb literature.  but this is my i-don't-know-much-about-this-topic-but-i'll-speculate-about-it-anyway answer.  as always: thoughts?