So, I was reading the GURPS thread, and part of a response struck a chord with me.  The response was [bolding mine]:

   I do not know what it is like for most people running the game, but I know in the case of my group (we play a few games/rules --Savage worlds, D&D, DCC, Mutant Epoch and GURPS) the players have by far the weakest grasp of the rules and how their characters can operate in GURPS  than any other system, despite having a play time advantage of 4 to 1 over any other ruleset. They have great fun while playing...but zero interest in reading even a little about how to best play their characters (options in combat and so forth can be extensive, but I mean even the most basic of things here) because the books are simply not engaging in the least. 

Now, I don't have many strong feelings about GURPS (played 1st edition when it came out in the mid-eighties briefly, but it never caught on with my group), so my thoughts weren't really about it as a system.  They were more about RPGs in general (maybe even OSR in terms of definitions), and specifically about the role of rules in an RPG.  I know there is often a dichotomy presented between "rules-light" and "rules-heavy" RPGs, but for some reason, neither description really encapsulates the games that engage my group.  We have games that should fall in either category that we do or don't like, so it's not the amount of rules that is the issue.  Instead, I think it is an issue that is expressed in the quote above.

Should RPG rules tell you what is possible, or should they tell you what happens when you try something?  That to me seems to be a big difference in how game rules are constructed (and, honestly, speaks to the difference between most types of games, like board games, and RPGs).  An RPG doesn't tell you what you can or can't do; it simply explains what the outcome of any attempt is.  I think that was one of the strengths of older editions of D&D, even with respect to its combat.  Look at the combat rules of the earliest editions.  Most combat was highly abstracted, primarily because it was not attempting to adjudicate every thrust or parry, but instead the result of 1 minute of fighting (or 10 seconds in BECMI).  So, it can be seen as the answer to the question, "I try to kill him.  What happens?"  I think this was one of the great strengths of the earlier editions.  The rules grew out of play, and specifically to give an answer to a player's attempt to do something not described in the rules.

I think that part of the difference now, especially in more modern game systems, is that the question becomes "What do the rules say I can do here?" as opposed to "I try to X... what happens?"  I don't think this is necessarily a conscious choice (and I'm willing to bet that there are groups, especially experienced ones, who maintain the freeform approach to actions even with very restrictive rules), but I do think that game systems can encourage or discourage this type of play.  I think one reason that feat or skill-based systems tend to bounce off of my group is that feats tend to limit players mentally ("I don't have X, therefore I probably can't do X or shouldn't try X").  Likewise, systems that involve "character builds" reinforce the same kinds of restrictions, mentally, if not in practice.

It's amusing to me that, despite entire systems being developed (and D&D being warped almost beyond recognition) to avoid having to ask the DM, "Mother, may I?", the basic interaction of many players with the rules system is exactly that.  I feel like the divide between systems where the rules specifically prescribe what actions are possible (sometimes in great detail) and those systems which just attempt to provide tools for adjudication is far greater than the number of rules (I think that it is completely possible to have a rules-heavy system that still has the descriptive focus.  I'd argue that AD&D is one... which is why many people want to argue that it isn't rules-heavy... because it doesn't play that way due to the rules only being necessary when a player attempts something, not to tell them what they can attempt).

To be clear, I'm not saying that one approach is "right" and the other wrong.  Obviously, this is a matter of taste.  But I do think it is a useful distinction.  It's one reason, I think, that despite trying to emulate earlier editions, 5e rubs so many players who like the feel of the earliest editions the wrong way.  Even though it has the potential, at most public tables I've played at the players are frantically looking for what the rules say they can do in the situation, instead of trying to figure out what they should do in the situation, and letting the DM figure it out.  5e says "rulings, not rules," but it most often doesn't play like that (at least outside my home table).  I wonder if the emphasis on universal resolution mechanics has had something to do with this (though, to be clear, I think it's completely possible to have a universal mechanic and still treat the rules as descriptive... I just think most games don't).

Should rules be primarily prescriptive or descriptive?

In most cases, rules should be both descriptive and prescriptive.  Though I'm not sure the descriptive/prescriptive breakdown is an exact match for the core bit, which is the player saying what they want to do and the GM adjudicating.  Everything else is commentary on that, whether presented as descriptive or prescriptive.

The very best prescriptive rules are open-ended enough to allow that adjudication while also serving as descriptive examples of how to do it and general guidelines for the most common activities that characters will try.  So it is easy for a "rule" to move between prescriptive rule and descriptive guidelines based on the experience of the GM and situation at the table.  That is also the most basic required thing that a GM decides, which is easy to learn to do at that basic level but requires experience to master.  Which is why when you get an influx of new gamers, there is also a heavy push towards prescriptive rules:  Some of them are struggling with learning how to adjudicate and a fair number of them don't want to be bothered and/or don't like it when you tell them it is the essential skill.

What the game should do becomes in part a question of goals.  Do you want the budding GM to run an OK game sooner, or do you want them to struggle now to ultimately do better than "OK" later?  And it is not a binary switch, but a slider, with pros and cons at every setting.  Me, I think limiting new GMs to simple constructs (say, dungeon crawls) so that they can get their feet under them with rules that have less training wheels on adjudication is the way to go, because a GM that doesn't start adjudicating early has to unlearn bad habits for the training wheels to come off. 

Should rules be primarily prescriptive or descriptive?

If I am writing a set of RPG rules and I want those rules to work the same an many different tables, I will want to write the rules in a prescriptive manner; "Take 2 actions from the following list of 5 action types, etc". This way anyone watching a YouTube session, or reading a actual play account will recognize the game as the one they play at their own table. The players can also get together and discuss various strategies and builds knowing that, for the most part, the characters will work the same regardless of the DM. Thus making an active gaming community around the game.

As a DM, however, I want the best experience at my own table and whether some group in California is playing it the same way I am is irrelevant. And to get the best experience, it helps if the players have more freedom to do unexpected things, thus a descriptive game is best for me.

So I don't think the descriptive/prescriptive divide is one of player preference, but rather the different design goals and needs of a game writer compared to those of a Dungeon Master. [Which is why I don't buy RPG rules anymore.]

GM's should express their settings to the players through the use of their system of choice to resolve tasks. The RPG rules *could* help describe that, and certainly the rules should "feel" right in mechanically expressing the "vibe" of the game and its genre.

CP2020 does a fantastic job of this. It's high-speed, lethal in its mechanical expressions of combat.

Other games might be more fiddly with bits and widgets to play with/against the system itself, as opposed to trying to abstract the actions of the game. Hyper-realism will do this, as well as games that rely overly on abstractions which pull you out of the actual moment with the PC's.

As a GM - you should be using the right tool for the job - and put your own emphasis on it. And your players are your audience, so you need to conduct your game to either 1) satisfy their needs, 2) convince them that your presentation of your needs transcends and includes their own.

I have had a lot more luck over the years with non-universal mechanics than what some are describing in this thread.

A Cleric uses a d20 to smash something with their club, throws 2d6 to turn a skeleton, or a d6 to open a stuck door. Who exactly has a problem with that?

I have gamed with countless dum dums, and they did just fine with multiple mechanics. I cannot see what the issue is.

Back to the OPs post, my OD&D/Basic players can get away a blank piece of paper to start their character sheet. If they need to roll, I tell them. Nine times out of ten, this happens after they have described their action. But, I will also say I find rules outlined in those books to be both prescriptive and descriptive.

What is lacking is the bloat, feats, and nonsense you referenced. That is where I think the nuance between prescriptive and descriptive likely becomes more important.

Streamlined mechanics are good.  Pushing that last bit of streamlining too far in order to get to universal is bad.  I find that a small handful of mechanics reused across multiple systems is better than "universal" and also better than trying to have a custom fit on every mechanic. 

That ties back to what I said the first time:  The best mechanics have some elements of prescription and description.  You don't get that if you shoehorn everything into one mechanic.  It tosses the description right out the window.  It's possible to keep both with every mechanic a custom fit for maximum description, but then the rules themselves get in the way of the new GM learning to adjudicate.

I’ll preface my own response with the fact that I have only limited experience being a GM or whatever a given system calls it, albeit some more as a player. I feel like as with anything else in gaming, it’s probably going to depend at least a little bit on what the GM and players want or have the most fun playing. For some groups they want to play more casually, “creatively” and rules-lite, in which case I’d probably lean towards recommending descriptive over prescriptive rules. In part because I feel like prescriptive rules handle a wider variety of situations well only if they really delve into the possibilities of said situations and the like, which may lead to a lot of rules that nobody remembers super well if they aren’t as invested in that kind of thing. But also, as a player with a preference for I guess maybe simulationist gaming (?), I have a liking for when the rules describe what happens and would sensibly and likely happen within a given scenario. As such, if it were a group of players and GMs like me, they’d probably love the idea of a super well designed and fleshed out prescriptive ruleset. (Admittedly in part because they’d spend more time occasionally in reading gamebooks and theorycrafting than in actually playing them.) That said, as Lunamancer had noted, any given set of rules can be modified to create different rules in play by the GM and their players. (Although the DM may have the most authority by the nature of the game and its possible unspoken or at times written rule 0, I feel like the players kinda have to be on board to some extent, as it is a social activity. Also because if they sign on to play a given campaign and rule system with a certain promised flavor, I feel it is likely to some degree reasonable for them to expect that to happen.) As they continued, it still helps to have a solid starting rules chassis and framework for innovation, and for that  you need some degree of prescription. So in a way, I’d have to concur that at least to some degree prescription is a duty of the original system makers. Likewise, they rely on GMs to adjudicate things that break the table’s immersion/suspension of disbelief/fun/whatever, don’t make any sense as written due to mistakes or oversights in the RAW, and to adapt the rules to more complex situations that require judgement calls, like setting the success number for certain skill checks relating to specific scenarios. Basically, to handle description, including narration. But at the same time, a good GM has to be mindful of the prescriptive rules to some extent, because those potentially matter to a scenario and how players will interact with and view it. Further, a good system will often have proper consideration of flavor and description. Or so I see it, anyway.