(Or how I posted in a discord and got told by the mods to write a blogpost instead).
In response to me looking to get some TTRPG people onto Roguelike Radio, a fellow discorder Markus posted a link to his blogpost on Bringing Roguelikes to the Tabletop - a considered and well written piece on how to leverage some of the design tropes from modern roguelikes into table top role-playing games. It turns out this is not the first time someone has thought of this idea: another discorder (SamR) pointed out that the Old School Renaissance was in part inspired by an rpg.net post where a succession of adventurers lived and died in a procedural generated dungeon. Markus' argument (made on discord, in response to me) is that the key three roguelike elements (procedural generation, permadeath and system interactions) have already been incorporated by the OSR movement, and so examining repeated runs and metaprogression is more worthwhile.
I'm not going to pretend that I have encyclopaedic knowledge of the OSR genre (I believe I own one OSR game), but this struck a chord, because Sersa Victory - designer of OSR game Victory Basic, has adapted Unexplored's Cyclic Dungeon Generation into a tabletop version. Unexplored takes inspiration from Brian Walker's Brogue in many ways, but cyclic dungeon generation is unique to that game and an interesting innovation in procedural generation techniques. Unfortunately the algorithm is brittle and prone to failure states at many points (such as destruction or loss of keys), and I believe simpler and more robust dungeon generation techniques - such as adopted by Spelunky and Brogue itself - are generally more successful.
And SamR goes on to point out the more obvious problem with procedural generation in TTRPGS:
Meanwhile, procedural generation has fallen largely out of favor, because there’s an abundance of bespoke modules and referees can create their own content... Like the implicit promise of procedural generation is endless content that is nevertheless comprehensible. But any GM can create that no problem. [Ellipsis added].
So if procedural generation doesn't suit table top play well* -- and I'm not going to argue against a presumably more experienced wisdom of crowds --, what should a tabletop roguelike look like?
Rather than look at the modern roguelike, titrated and diluted through decades of implementation, I'm going to go back to our ur-text: Rogue, itself, to try to find some design principles that can define what a tabletop roguelike could be.
The first is a straight forward response to SamR's very excellent point. Glenn R. Wichman, Michael Toy, Ken Arnold and Jon Lane created Rogue because they didn't want to have to DM (whatever that mean - direct message? maybe email was too slow) and just wanted to play. A video game allows a set of constrained systems and rules to (mostly) safely operate for the player; whereas the explosive medium of table top play rules requires some kind of dampening effect like a carbon rod being inserted into the fissile material in a reactor -- a game moderator (GM) if you will (this is not a metaphor for sex maybe?). But this sounds like a lot of work and not enough play for one person (I suspect they should get paid). So any roguelike game is similarly going to forgo the need for this game moderator to allow everyone to play instead.
So a tabletop roguelike must be:
1. GM-less.
The second constraint naturally flows from the first. In a shared world, where everyone is a player, there is no authorial fiat: everyone must agree to the actions, outcomes and events (the systems). This is relatively straightforward in the game present because we can constrain situations of play to be within a specific space and time, but when it comes to backstory: it is too easy to create and miss contradictions in the wider world being made. So much like our Rogue, we are going to descend into the dungeon with a clear endpoint in mind (The Amulet of Yendor, in the original) and free of any past that might complicate this. The systems within the game may allow this past to be constructed, but we certainly don't want to play part way, and then realise we missed an important fact in a spin-off novel that negated everything we did.
2. No lore, one goal
(The easiest way to avoid the need for lore is to set the game in a modern setting -- any well documented historical setting would also suffice. The fantasy settings of OSR games might lean on D&D tropes enough to also apply -- certainly Rogue depends somewhat on this).
The third constraint will be where all the load-bearing design sits.
3. You can learn from your mistakes
You, as in you the player, not you the character. This is has many implications: the game systems must be complex enough that a simple solution is not obvious or (equally) random, but for which your gradual comprehension of the underlying rules and your capabilities as a player are of benefit. This also implies the possibility of an auspicious outcome instead of a steady descent into failure (even if auspicious simply means the last one to die, or the one to die most memorably). This also implies some repetition of systems, either through repeated play, or through enough similar scenarios that repetition dominates (but does not eradicate) variation.
Now I have sketched out the three design principles, it is relatively easy to cast around and find games in the same mold. I'll just pick three out:
Decaying Orbit - https://storybrewersroleplaying.com/decaying-orbit/
The Zone - https://thezonerpg.com/
Last Train to Breman - https://seaexcursion.itch.io/bremen
You'll notice pretty quickly that two out of three of these examples have at least one modern roguelike feature:
4. Cards
---
* I can think of two examples where procedural generation can still be successfully accommodated in GMed TTRPGs: scenario creation and character creation -- particularly life-path style character creation. In both instances the participant is essentially playing a solitaire game with its own rules and constraints that then feeds into the larger shared play.