Well, here is the problem - "smart play" is largely defined by what gets a lot of bang for your buck under the rules. You don't choose to buy a notebook full of scrolls until *after* you realize that the rules make it possible, and that it is very economical.
I think that we should differentiate between natural play and smart play. Natural play is the stuff you'd try to do just because it makes sense in the fictional framing. "I draw my sword and try to stab it," is entirely natural, an common thing someone would try to do. "I go out and buy three dozen scrolls," isn't really something you expect folks to try to do, just as a part of generic fantasy. I find it an issue if a bit of natural play is really broken, and I'm less concerned with play that comes after attempts to optimize are broken.
And here you're creating a semi-arbitrary distinction.
But even if we go with your distinction, I will absolutely say "I try to buy the most cost-effective equipment I can afford" and "I try to learn and have ready as many spells as possible" are 100% expressions of natural play.
See my Wand of Cure Light Wounds example. If that thing is on the market at less than half the price of a +1 sword it is highly un-natural play to not try and pick some up (or one of the alternatives) for the party. It's lots of healing in one little happy stick at a bargain price. The 3.0 economy once again falls to natural play with just a very little knowledge of the world.
And then there's the saving throw system and the way Save or Suck and Save or Die trumps hit point damage. Natural play for a wizard is to want spells that will make the enemy stop hitting you.
Why didn't it get picked up? Because of learned unnatural play due to Gygax and Arneson's system. Where Save vs Spell (i.e. Fireball) was much much harder than save vs Poison (i.e. Stinking Cloud). If you play 3.0 using the unnatural tactics picked up from 2E it actually works pretty well. Come into it cold and try natural tactics with the economy presented, the bear druid, or wizards who try to prevent the enemy hitting back and it breaks accidentally.
I think this is a "hindsight is 20/20" thing. Today, in an age where we have RPG playtests that have tens of thousands of people, "testing" means something very different than what it did in 2000. Or, perhaps a bit of armchair QA-analysis. It is very easy, now, to say they *should have* caught it. But, have you ever tried to really test an entire RPG system, with only a handful of people?
Do I have the budget for 3.0?
And the simple answer is
Gygax did. Admittedly oD&D is the best playtested game in the history of RPGs. But that's because Gygax learned his playtesting from and with wargamers and wasn't afraid to de facto errata. oD&D was playtested by wargamers playing to win. There are
few RPG designers
Taking a cue from software development - you will generally only catch failures for which you have an explicit test case. If, during the playtests, nobody really thought of trying that action - it was not *natural* play - then, yeah, they'd miss it.
And taking a cue from software development
you send your testers out to try to break the software. This is what the 3.0 playtesters simply didn't do. 3.0 works on the use case expected by the designers - which involves carrying forward a lot of learned behaviour patterns from 2e.
Yeah. Note how 13th Age had the opportunity to learn from 3.0's mistakes?
Note how e.g. Marvel Super Heroes didn't. And it doesn't have those problems either.