In working on Dust to Dust, one of the significant areas of design theory discussion has been which rules to reveal and which to keep secret from the publicly-posted rules. That secret rules exist should be no surprise to most gamers; in one form or another, it's a practice as old as the roleplaying hobby. I've been trying to think of an example of secret rules outside of roleplaying games or roleplaying-derived games, and I eventually came to a Wikipedia page dedicated to that very idea.
Let me get one thing out of the way. There is most assuredly a wrong way to design secret rules. You do not want your game's rules to feel like this or this.
What am I talking about with secret rules in roleplaying games? Well, going all the way back to original D&D, there was a concept that only the DM should read the DMG. Up until 4e's publication, the rules for magic items had always been published entirely in the DMG, and moving those rules to the Player's Handbook was a controversial decision among the fans. The DMG also has all of the rules for NPC morale, building strongholds, monster abilities, traps, and so on. These rules are kept secret from the players on the principle that discovery and exploration are key emotional rewards of player fun. It also preserves the DM's right to do anything he damn well pleases, and (if secrecy has been maintained) the players won't be able to say much about it. Of course, this becomes a complete fiction as soon as another player in the group steps up to run a game. In my experience of DMing 3.x, the only truly secret rules were things I wrote myself. Otherwise, players might be unfamiliar with a rule published in a book they didn't own (The Really Definitely Complete This Time Warrior), but I would not have considered refusing to hand over that book on request, so it was more "lightly obscured" than secret.
Setting-specific material (such as, say, regional sourcebooks for Forgotten Realms) presented themselves as having actual secrets for the DM, and much less in the way of rules that players would ever need to reference. These for the most part stayed secretive by group consent, though this too fell away over time. Other game systems made a stronger case for secrecy in their early release - Deadlands comes to mind, as the book was split into "what people think is going on" and "what's actually going on." I'm not sure how much of this really came down to secret rules, though; perhaps one of my readers will chime in.
Which brings me, circuitously, to LARP design and secret rules. LARPs publish a rulebook that generally tells players what their characters can learn to do and how to react to the various awful things that monsters do to them. The pace and flow of LARPing means that designers need to reduce as much as possible the number of taglines players have never heard of before, so that every fight doesn't start with a briefing on tagline rules. I say this advisedly; Dust to Dust has a sort of famously large Effect List, but we really did try to include every tagline we intended to use, knowing that the rarer ones would require pregame reminders as to their effects. LARPs may not tell you how a monster will use an effect, but they do have to be transparent about what the effect is in ways that tabletop games don't. Even monster stats often can't stay secret for long, as many games require players to spend a few hours of each event playing monsters or other NPCs. Sportsmanship and confidentiality are key, but the surprise is ruined for at least some portion of the playerbase.
And now, I conceal the rest of the discussion from you.
(No, not really.)
The rules that we bother to conceal in LARPing are therefore different. The baseline good secret rule is one that represents an in-play discovery, such as a character learning a heretofore-unknown power. For example, Shattered Isles introduced bonds as a kind of power set, but initially released them as minimalist accents to a character's power; with only three or four powers on the list, one would be hard-pressed to build an effective character on those powers alone, and the design didn't intend that as a possibility. I was new to the game and not in contact with the design staff at this period of the game's history, so I don't know what was intended from the start as opposed to becoming intended along the way. Perhaps another of my readers will speak to this point.
In the fullness of time, however, the game revealed that the published powers were just the tip of the iceberg. Each bond had two more ranks of power (Intermediate and Master), and eventually a Lost Path and two Related Powers. This opened up a lot of room for all-new powers and upgraded forms of existing powers. It added a lot of exciting discovery to the game for a lot of players. It allowed designers to explore themes within the elements and Spirit that they had not gotten to fully explore in the schools of magic, given that almost all bond powers were per-day and self-only. It was, on the whole, a good design space to open up.
This approach was broadened in King's Gate to include the six totemic bonds and the martial schools. In all cases, only the Basic levels of bonds and martial schools were publicly available, and the rest were revealed over the course of play. Importantly, these effects were put together from other familiar effects, or were effects that only the power's user needed to understand. In King's Gate, the bond powers were still typically not a character's whole power set, as only a few bonds even had something that could be called an attack, and such powers were very expensive to purchase beyond three uses per day. Martial schools were much more frequently a character's entire (or nearly entire) power set... given that advancement within the martial school required advancement within the Wounding Blow progression.
Eclipse continued this pattern, initially revealing only the basic levels of its Combat Disciplines, Mutations, and Cybernetics. This had some unintended outcomes that, at least from my player perspective, were not preferable, and they have since revealed all grades of standard powers. The problems I had with this had to do with the feeling that the second and third tier of powers in some paths revealed those build choices to be objectively superior at the thing I had wanted my character to do well, in ways that were not indicated or strongly hinted in the first tier of powers.
The specific case of Eclipse highlights a problem that I have with secret rules of this kind in general. SI, KG, and Eclipse asked characters to make immutable character decisions from the start of play. Bonds, mutations, and cybernetics have been treated as nigh-impossible to acquire after character creation. This gives an overwhelming game advantage to experienced players, who know what to expect from various power sources and can plan accordingly. First-time players, particularly those who are not already well-connected in the game's community, have a much harder time of it. Eclipse also has Seekrit psionic orders, but these are for the most part not locked out by decisions made at start of play.
As a counter-example, the Wildlands campaign asks players to make very few choices at character creation that lock them out of future options, aside from their choice of race and class. Wildlands has a long and involved history of Lost Arts that require the player to find someone with that skill and the Instruction skill in order to learn. Many of the most important Lost Arts (affectionately known as "Misplaced Arts"), such as Dexterity Armor, are advertised in the campaign book. The system also has a number of race-change and Spirit Forge options.
The thing that I think is kind of strange about the through-line trends of SI, KG, and Eclipse is that no concept of secrecy spread to other parts of the game. In SI and KG, players had to put forth a certain amount of effort to collect all of the brew recipes listed in the rulebook, but there was no plan in advance (as far as I know) to keep a certain list of brews out of the rulebook and reveal their existence through research. Voldari/Brewer's Inferno in SI/KG is the clear exception here, and I believe there were a few mega-poisons that were intended from the beginning of KG's brewing plot. Instead, the game had to restrict research rather tightly, as the Brewing skill was balanced on the assumption that the list published in the rulebook was complete. Likewise the schools of magic, though the exceptions there are interesting stories and somewhat more numerous. A certain amount of research was permitted within Ballistics and, later, Mechanics, but I'm similarly unclear on which parts were always intended to be new discoveries by PCs and which parts were added when suggested by a PC. In Eclipse, some research has been done within Med Tech and other fields, but here again it feels like the things that are introduced are rules tweaks because Plot and Rules agreed that something new could fit in there.
Okay, in my experience, players love to research new things. It gives them something cool to do with BGAs, and the reward is fantastic: a spell or whatever that you can share out to your friends and sell dearly to everyone else, since you're the only one who has it. Why not build research into the rules from the ground up? So that's what we've explicitly done in Dust to Dust. Ritualism, brewing/Alchemy, Inscription, and Forge Magic have listed in the rulebook a limited set of their effects. For example, the Ritualism chapter shows twenty spells, of which players can pick several for their starting spellbook. This is trivial, however, compared to the number of spells currently created. We've shown a handful of those hidden spells in the hands of the powerful wizards of the first and second (NPC-only) World Events. The same is true of the other forms of magic, though to a lesser degree. We're expecting that these new formulas will enter play through research, loot, and trade, so that each player who seriously pursues those arts always has something to chase. The text attached to each formula even gives hints as to what your next goal might be.
This doesn't mean, of course, that we can't expand the rules with a new-to-us formula. But the odds that we'll need to are much reduced, and even research projects that are pursuing something we know to be doomed to failure can turn up relevant and useful information or new material.
I've often said in the past that I hate secret rules. This is only half true. What I hate is making permanent decisions on partial information all on an out-of-play level, and because I don't like it, I assume that other people dislike it. On the other hand, I love exploration, and (as stated above) I think this is universal. Now that I come to the point of it in this post, the difference between the two is a little tricky to verbalize. Because players will be spending XP on Ritualism (or whichever skill) rather than on specific effects, newcomers shouldn't have gross misconceptions about how much the skill has to offer. (At least, not to the degree that someone who only saw the basic level of a bond would be mistaken about the overall power of a bond.) There's a lot of depth to explore, and the spells in the rulebook aren't really indicative of the power of ritualism as a whole, but there should be enough shown and hinted at to give even new players a pretty good idea of what the choice represents.
Secondly, because players can initiate in a school of magic long after character creation for an increased cost, they only lock themselves out of ritualism if they choose to play a homunculus (and then it's only a partial lockout). The other highly expansible arts have no restriction on post-creation purchase. The power sets that do not encourage expansion - such as Warrior Orders or Celestial magic - have much less in the way of hidden material.