Dimensions / rules? - signal to noise - if 1 of total 60 bushes has a secret, you'll lose to STN. Bash a few - learn they don't have secrets. Do NOT mix your messages. E.g. character barks as well. - diegesis - is a "collection" diegetic or not? Not sure what implication this has - substitutability - can you swap one thing for another? E.g. when collecting fish - fishing is the same i.e. the process, it's just the outcome that changes. So collecting 5/10 is v similar to 10/10. Contrast with fun handcrafted puzzles - asking for 5/10 leaves content on the table. - clues themselves - how do you signpost info, and what info? Deducibility matters - no "point n click adventure game logic" --- This topic first occurred to me _as_ a topic in hearing about my wife's exploits in _Stardew Valley_. I can't recall which aspects of exploration/discovery/hidden treasures had proved somewhat frustrating, but that was the rub: something which in principle should have been rewarding (exploring for secrets) was not. On reflection, this is hardly limited to _Stardew Valley_. Thinking back through many of the games I've played, a majority are sprawling games with large worlds or sandboxes, which heavily select for this very problem. These reveal maybe the first two principles of game design we'll think about, which basically motivate the _use_ of secrets: 1) The explore-discover loop (let's call this the EDL) is motivating for players. This is a particular instantiation of a feedback loop: some process (exploration) leads to some kind of reward (discovery). We'll want to characterise the process - or _mechanism -_ as well as the reward itself. Possible theories of **mechanism**: - Intermittent reward: the EDL is in some ways just a Skinner box - taking a step into a screen you haven't seen before is, under this theory, isomorphic to opening a lootbox. Will it contain something interesting? This is important to consider because it would have different prescriptions: if intermittent reward is specifically what we're after, _not all screens should contain a reward_, and you actually in some ways _don't want it to be deterministic_ - leaning more on chance than on pure skill or puzzle-solving. This would be a more "intrinsic" model of the motivation. - Catharsis: the EDL sets up a challenge (exploration) that, in its overcoming, creates catharsis (discovery). This may have several flavours - it could be more or less skill-based, versus grinding-based or something else. However, it probably lends itself more towards determinism, though not necessarily - the challenge could be grinding, with a high expected time-to-discovery, which might look more like a lootbox though the core of the satisfaction lay in overcoming the challenge. - ? Others Possible theories of **reward**: - Instrumental: the reward is progress against some other (terminal) goal the player actually cares about. Walking around and finding hidden caches of money might be dull moment-to-moment gameplay, but the uses for that money could be so rewarding that this actually creates an engaging EDL. - Terminal: the reward _is_ the thing that the player wants to get out of the game. At one extreme you would have something like the discovery being a satisfying piece of narrative - something like environmental storytelling - which doesn't contribute to anything outside of that moment (nothing else is changed by discovering it, modulo the accumulation of such narrative moments the player's understanding of the rest of the game). At the other extreme, the game is explicitly about dopamine, and the discovery produces a lot of flashy effects that are basically what you're there for and otherwise doesn't produce instrumental value against some longer-term goal. Maybe pure point-scoring could be an example; yes there is instrumental value in getting closer to a high score, but the scoring of points may be satisfying in and of itself. - Blended: I think most instances will have components of both. For instance: exploration leads to completing some sidequest that is both well-written (producing terminal value through narrative) and awards experience (producing instrumental value through progression in an 'outer' gameplay loop). -- Some examples we can reason about: **The (Platonic) Ubisoft Sandbox** - Morsels are sprinkled across the game surface, but everything is marked. - Trades serendipity and incentivising open exploration for clarity/transparency. - Any surprise comes from the environment in which the morsel is found (could be surprisingly scenic, etc.) or from the specific detail of the morsel (e.g. it's marked as an XYZ category of morsel, but it turns out to have a specific bonus you care about). - Trades 'skill' in discovery for 'satisfaction' in completion. **Minecrafty Procedural Generation** - Can draw on meta-level skills in parsing the generation (though this can be shortcut with tools that do it for you) - More Skinner box - Assists you in your journey but otherwise not really significant nor "curated"; random instrumental value, no terminal value.