Speaking from my experience here, there are a couple of things I would recommend:
- DM Cheat sheet - this was really helpful for me, since it gave me an idea of the basics of the character. I made sure the player created and provided me with this sheet at every level, not only to keep me informed with what their character was doing, but also to give me insight into what they thought was important about their character. This helped me to build obstacles that catered to their character's specific role, or at the very least what the PC perceived their role to be in the party. I would limit this cheat sheet to 1/2 a page at lower levels, and 1 page at higher levels, to ensure it remains concise and useful to the DM.
- Time Limit on turns - this was a more controversial move in our group, but definitely helped keep my PCs on top of their characters. If I as a DM had a question, or asked their character to do something, the complicated PC would have no more than a minute (or 2 depending on how complex the request was) to figure out the answer. By having the PCs aware that this time limit was in effect, it incentivized them to maintain a deep understanding of their character. Our table also had a high level of trust, so I wasn't too worried about my PCs fudging their characters in these time sensitive scenarios.
- In advance verification - this was something I only used once, but can still highly recommend. Before allowing my PC to create a complex, multi-classed character, I sat down with them and discussed why they wanted to create such a character, and how we could work together to streamline the process in-game (this was incidentally how we landed on the cheat sheet solution above). By having this talk in advance, you can understand why the player wants a convoluted character, and brainstorm alternatives if possible (eg Player: I want to be a healer, but also be able to fight. Me: well instead of multi-classing and trying to use 3 prestige classes, you could just main as a Cleric, and by taking feats X, Y, and Z still achieve this goal). If alternatives are not an option (eg Player: No, I really want to play [insert convoluted character]!) then at least you've now gotten some perspective on where they are coming from, why they have chosen this particular build, and laid the groundwork for future discussions should things go off the rail.
Introducing a metagaming requirement is an interesting idea, and something I had not considered. While this would be useful in the context of the game's lore and the character's development, if you are more concerned about how to deal with complicated characters as a DM, I would just warn you that complicated backstories can lead to complex character motivations, which in my experience, often leads into My Guy problems.
That being said, the flip side is that you are maximizing the number of plot hook you as a DM have available. Is the character adamant on being a bard/druid/green whisperer/sublime chord/arcane hierophant? Fine, but their complex background has made them a target of a group of purist druids, who believe their magics are being abused by an outsider for their own personal gain.
Ultimately, requiring as comprehensive a backstory as the character is complicated can work both ways, but independent of this, I would advise you to consider some of my suggestions listed above. They can definitely streamline the DM process, and make your life a lot easier.
So, I don't read the rules the same way you are. It seems to me that establishing a hold is a grapple check, just a certain specific kind of grapple check, and so these abilities that make you larger for grapple checks make you larger for that purpose as well. In any case, I think it's clear that the authors thought it did so: that there is no benefit anywhere listed that explicitly increases one's size for the purpose of establishing a hold is telling.
Powerful Build (and the similarly-worded "Larger than Small" class feature) are clearly meant to make you count as larger for basically everything combat-maneuver-y (combat maneuvers aren't a defined thing in 3.5 but "subject to a size modifier or special size modifier for an opposed check (such as during grapple checks, bull rush attempts, and trip attempts)" is pretty clearly trying to evoke that sort of stuff, as opposed to stuff like Intimidate checks or whatever that might also work out as side benefits), and grappling is so exemplarily combat maneuver-y that it's even called out specifically in the parenthetical in most of these upgrades.
Basically, there's nothing other than straight up size-alteration that will let you do this, but I think that's because the designers thought they already had that covered (and I would argue they do already have it covered), rather than because they actually wanted small creatures to really never-ever-ever be able to initiate grapples with bigger opponents.
That said, even with this restrictive interpretation of the text, we can still grapple enormous opponents quite handily, if we can provoke them:
The step required to establish a hold only applies when one is seeking to initiate a grapple. That means the size restriction can be ignored in the prominent case of an opponent having already initiated the grapple.
The Tarrasque is built to grapple its opponents: it possesses Improved Grab, Swallow Whole, and a +81 grapple modifier. It shouldn't be terribly difficult to get it to initiate a grapple with your halfling without any effort on your part, but suggestion et. al would certainly be effective should the Tarrasque prove hesitant to bite.
Once the grapple is initiated, your halfling can pin the Tarraque, move it about, damage it, prevent it from leaving the grapple, and all that good stuff without having to worry about size limitations. The size limitations are only a problem against opponents the halfling can't taunt into initiating a grapple.
Best Answer
Yes. It doesn’t even require especially dubious interpretation, just no one deciding enough is enough and putting a stop to it. And the basic combo even fits in a “Core + 1 book” environment—this really doesn’t require a ton of dubious interactions between disparate parts.
The key is the manipulate form ability of sarrukhs from Serpent Kingdoms. Serpent Kingdoms is a Forgotten Realms book, and it only works on a “scaled one” native to Toril, so those are “issues” here, but neither requirement is really a significant hindrance, speaking hypothetically: we can just assume we’re talking about an FR game—which means everyone is presumably native to Toril—, and plenty of PC-eligible races count as “Scaled Ones.”
Anyway, manipulate form allows a sarrukh to raise any stat of a “scaled one of Toril” up to any number less than or equal to their own value in that stat, as well as give that creature “an extraordinary, supernatural, or spell-like ability,” with no listed limitations.
Here, we see three major limitations:
It’s supernatural, which makes it unavailable through many forms of polymorph magic.
It cannot be used on a sarrukh, which means a sarrukh cannot use it on itself.
It cannot increase a stat beyond the sarrukh’s own stats, which means it can’t get arbitrarily large
But these limitations are all trivially obviated:
While most forms of polymorph magic do not allow you to get supernatural abilities, there are plenty that do, e.g. the core 9th-level spell shapechange.
While you cannot use manipulate form on yourself while you are sarrukh, and therefore require a partner, many spellcasters get a familiar, who can be presumed to be absolutely loyal, and shares in your spells. Furthermore, there’s a core familiar option—the viper—that counts as a scaled one.
While the sarrukh is limited to its own scores as a maximum for how high it can pump an ability score, there is no limit on how or why the ability score is what it is. That means the sarrukh can use bonuses to increase their score, and then increase their partner’s score to the same value—as an actual, base value, without any bonuses.
So any scaled one of Toril who has a scaled one familiar (e.g. a viper) and can cast shapechange can share that spell with the familiar, and then for the duration of that spell one can be a sarrukh and the other can be the scaled one subject, and then they can swap. In between, bonuses can be applied to the one in sarrukh form to increase its maximum values, so that when it uses manipulate form it can increase its partners values higher than its own—and its partner will be able to add another increase on their own turn.
For example,
And that’s not even getting into the question of other abilities—on which there isn’t even the limitation of the sarrukh itself having the ability in question. As written, there isn’t even any limitation that the ability be one that actually exists in the written game—though even most theoretical optimizers balk at that—and surely any ability that does exist and Pun-Pun can become familiar with via, say, shapechange, is fair game. But we could easily point out that, ya know, manipulate form is itself a supernatural ability. And there are extraordinary abilities that increase ability scores, so they could just give each other a bunch of those instead of bothering with the swapped belt of giant strength. On some level, the above approach is taking some extremely conservative interpretations of manipulate form, as absurd as that is.
Anyway, none of this is at all dubious from the perspective of the rules as written. All of it is completely straightforward applications of manipulate form, just used by a coordinated pair of creatures who, together, can work around its few limitations. Serpent Kingdoms was a notoriously poorly-considered book; the sarrukh and its manipulate form are far from the only absurdities in it (though it is by-far the worst of them). Earlier I said that the main limitation on manipulate form was that the sarrukh could not use it on themselves, but that’s not really accurate. The apparent thing the authors expected to prevent this nonsense is the fact that the sarrukh is an unplayable monster. As they so often did, they neglected to consider the possibility of a player getting access to monster abilities, i.e. manipulate form.
Now, various approaches to Pun-Pun do involve more tricks to allow you to pull the combo at lower and lower levels, and to get more abilities that might otherwise be inaccessible (e.g. class features that no monster has, or abilities that were defined but never given to any creature or class, or abilities that are just made up on the spot), and so on. Some of those tricks are somewhat more dubious, RAW-wise, than the basic combo.
One of the most famous renditions of Pun-Pun (the first one to work at 1st level) involves Pun-Pun as a kobold paladin who gets a wish off the demon prince Pazuzu, because you see, the description of Pazuzu says that the first wish makes someone Chaotic, and the second wish makes them Evil. Because Pazuzu delights in pulling down paladins, he always endeavors to give paladins everything they want with the first wish, to make it all go perfectly well for them, to try to encourage that second wish. One wish is all that Pun-Pun needs to get the combo going, though, and so the advantage here is you get one absolutely-untwisted wish. But this relies on an NPC description being absolutely iron-clad, on Pazuzu actually being a thing in your setting, and also on a 1st-level paladin knowing about this. People have some quibbles with some or all of that. But other ways of accomplishing 1st-level Pun-Pun, no Pazuzu needed, have been found since then. Even if you disallow all of those, if nothing else, a 17th-level character can just cast shapechange themselves, no wish required.
(For my money, though, Pun-Pun will always be a 1st-level kobold ex-paladin.)
So yeah, one way or another, Pun-Pun is 100% RAW-legal. It’s a fun little testament to the absurdity of the game, an occasionally-useful trivial answer to optimization questions (to illustrate the need for boundaries and constraints on the question), and possibly a challenging problem to work, to optimize it more and get the combo faster or more safely.
Pun-Pun is not playable. No one involved in figuring the combo out has ever recommended playing. In fact, I strongly doubt anyone ever has; it’s just pointless. Once you have achieved Pun-Pun ascension, you can literally do anything you want, with guaranteed success, with absolutely no concern for rules like turns or actions or resources. So even if it does happen it’s kind of like “OK, that’s cute; we’re going to ignore that you did that, and keep playing our game now, because we really don’t appreciate you summarily ending it on us like that.”