Redesign Easy Mode

[tmewett]

I think we should have a chat about what you see the purpose of Easy Mode being. A track for "onboarding" beginners to real Brogue, or a different yet satisfying experience in itself, etc. Different only in balancing, or in content, or game structure.

Brogue's Easy Mode is where you take less damage but also get less points, right? I'm quite not sure why, but playing it just doesn't feel like Brogue to me. I'm not making any of the same tactical decisions. I can just easily steamroll everything - until I can't. (Also, players who can't beat the game anyway don't care about score.) In contrast, when playing with auto-ID, while I'm not making any of the same strategic use-ID decisions, I am making many of the same tactical decisions (on when to use a specific item, which enemies to avoid, and so on). And those are what I find interesting.

So I guess Easy Mode should be in some way interesting to play in itself, perhaps by having most of the same types of decisions being made as the base game, but with a greater tolerance for player error in making those decisions. I don't know if that could be better achieved by spawning smaller groups of enemies, for example, than by reducing damage levels.

@HomebrewHomunculus in this thread

Possibilities for this:

• spawn smaller groups of enemies
• spawn some % of items identified or magic-detected
• start with more/better gear

Personally I like the identified items idea. New players get presented with a whole bunch of items with unknown powers, and it might take many hours to really work out what everything does. If they can play an easy mode where they can pick up some cool magic right at the start, read what it does and start using it, I think it would be more interesting.

If we change game generation, we'd likely need to change the mode itself to not be a toggle at an arbitrary point and something you select from game start (new menu entry?)

[withinwheels]

A related idea that I've been thinking about off and on is a difficulty filter for seeds. We're able to examine (say) the number and placement of enchantment scrolls, food, consumables, positive runics, etc in a seed before presenting it to the player, and I suspect we could reduce that to a somewhat meaningful "difficulty score" without too much effort.

"Easy mode" might then be handing the player a seed with a low difficulty score. Instead of modifying how the dungeon is generated, we merely pick one that looks more tractable. It has the advantage of still being a fairly representative experience (nothing is 'dumbed down"), and it might still be mixed with other ideas like reducing pack sizes.

[epmoyer]

I very much like the guiding principle of "playing in easy mode shouldn't teach you bad habits".

Playing 1.7.5 in easy mode (which I did a lot of) was good in some ways: I learned about monster types, item types, the warden, and the mechanics of what actually happens when you grab the amulet. But it gave me an incorrect "feel" for enemy difficulty and I had to completely re-learn their actual difficulty after switching back to regular mode.

I like the idea of selecting the difficulty mode at start; doing so opens a world of pre-processing possibilities and personally I never switched modes mid-game anyway, so I think making that choice brings way more value than it takes away.

In a perfect world, I would love to have multiple levels of difficulty. Today Brogue is a game that you have to loose hundreds of times before you ever win, and while purists might argue that is part of its charm (and part of the charm of roguelikes in general), it's entirely conceivable to imagine a world where you loose a bunch of times on the easiest difficulty, then finally win a few, then crank it up a notch.

I love the idea of partial identification, and would also welcome an additional difficulty setting with full identification. Brogue Lite does that today, so apparently there is a "market" for people who would like to play in that way. For CE I'd argue that no items be removed (whereas Lite removes a bunch of things like negative scrolls, Hallucination, etc.). Why keep them? Because if "easy" modes are there to teach you how to play the "real" game, then you should see (and be able to play with) all items. Leaving them in lets you both learn how they work, learn their statistical distribution, and learn where they may or may not appear.

A thing I'd love to see (though I realize the implementation is challenging) is a way to alter difficulty (and play time) by "compressing" content into fewer levels. Coffee Brogue attempts to do that by spawning more valuable items on early levels and having down-stairs skip a few levels randomly. It would be really cool if a skilled player could select a "quick" 6-8 level game and play through it in 30 minutes to an hour. I can think lots of reasons why compressing 26 levels into, say 8, is hard to implement / silly / broken, but what IS true is that IF it existed, I would sure as heck love to play it, both as a "training" experience and as a "short pick-up game" experience.

In the interest of taking an "Agile" approach to what the "MVP" of an easy-mode change would be, It seems very attractive to call that:

• Adding a mode select at game-start
• Implementing partial (and possibly also full) identification
• Filtering seeds by some scoring system (per @withinwheels suggestion above)
• Removing the current in-game easy/hard toggle

The first three bullets get implemented at start-time only, so none of the core logic has to change. That seems like a great way to minimize the risk of breaking the core gameplay. The fourth affects the core, but is very compartmentalized so the risk of breaking the core with that change seems low.

[epmoyer]

Now that I've gone off to think a bit, I'm going to argue out-loud against my third bullet above (Filtering seeds). The attraction of filtering is that it produces a 100% "normal" level that the player might otherwise have encountered on normal mode (it's just a regular seed like any other). But the downside is that the implementation is hard to develop and potentially very time consuming to execute. One needs to iteratively "render" random seeds and evaluate each against some scoring criteria. The time to find a seed matching the scoring threshold is theoretically unbounded, but moreover the scoring criteria (for developers) seems difficult to communicate/audit/evaluate/refine over time. Contrast that with the challenge of writing an algorithm that begins with a given random seed and then post-modifies the level data to make the levels "easier". In practice, that algorithm would (I believe) be simpler to implement.

• It is deterministic in time. 1 pass; just render the level data normally, then make additional modifications if "easy" mode is enabled.
• The algorithm is easy to communicate. It's going to end up looking something like: Sprinkle qty x of items from list y on level(s) z-q, Remove curses from items according to some straightforward randomization, Remove some monsters according to some straightforward randomization, etc. Buff some items according to some straightforward randomization. Add some food. Etc. Seems like when one is done, one will be able to summarize it in 8-10 bullets that any player (or dev) could understand.
• You can develop it in pieces (e.g. Add items, test/try. Add food, test/try. Add buffs, test/try).
• Making a change at the 11th hour before a release is low risk; you don't have to go back and rebalance a complex scoring algorithm; adding/removing items from the "drop bag" lists, or tweaking drop/buff percentages is very low risk. You might get the balance wrong, but it's unlikely that you break the game.
• It seems relatively straightforward to debug.

[HomebrewHomunculus]

I've been thinking about this.

There's a few different angles to approach the issue. I was going to talk about the variant gameplay systems vs. variant generation parameters. But I actually think game length is a really interesting thing to think about in relation to easy mode/beginner-friendliness/"player onboarding", and not commonly discussed, so I ended up writing a lot about that. (Probably too much, actually - apologies.)

Variant gameplay systems (soft-ID) and level generation parameters (smaller hordes) will definitely need to be an up-front selection in the menu, I agree. A variable game length is possible to do either as an up-front "hard" setting in the menu (which would affect the level generation parameters) but also as a "soft" setting in which the player has a choice, during the game, to take their winnings and run. Of course, Brogue already has a soft choice of game length, in some sense, in being able to either escape with the Amulet of Yendor, or push beyond for the Lumenstones.

Being able to end the game earlier does not truly change the game or make it easier, of course, but I think it can have an effect on how the difficulty feels to new players. Many roguelikes feel so daunting, I think, partially because any measurable success can take so long to achieve, both in terms of total playtime and the run length.

Game length, zone structure, and memorable deaths

If we take a hypothetical game that will take 1000 hours for a player to get their first win in, then that's a very long time to wait before a player gets proper feedback that they've mastered the game. Between 100 and 500 hours, the player will likely have got the hang of things, but spends a lot of time dying in the middle depths. They might be slowly learning various advanced strategies, but due to the high variance in the depth that they die on, it's hard to observe this improvement. If a game has 50 levels, then it's hard to see if a death at level 33 or level 38 had any meaningful difference between them.

I think Brogue has much less of a problem with game length than most other roguelikes, certainly. But that doesn't mean that the feeling of working towards it couldn't be improved. Dying at level 11 or 14 will not feel like very different results, I would say. Especially in a game with a single dungeon without have very clean-cut "zones" of levels.

Zones as tests of subsets of skills

I definitely think that Brogue has good variety and progression in the dungeon contents. As a player I can often tell that I've reached new depths even without checking the number, bit just from the feel of the different sub-zones, key holders, monster types, and so on. Still, that cpmtemt progress happens sort of gradually, so there is a slight "frog in boiling water" effect.

The content progression in Brogue is not clear-cut like in e.g. Spelunky, where it's very structured and fixed: 4 levels of mines, 4 levels of jungle, 4 levels of ice, etc. The game doesn't just vary the palette of things placed within the dungeon, it changes the palette completely: different tileset, enemy set, and traps. (I think the only thing that stays constant throughout is the item drop table for shops and crates - any weapon can spawn on any level.) For example: in Brogue, goblin warrens are one distinct sub-section of a level that can spawn on depths 5-15. Whereas in Spelunky, a goblin warren might be a full level in itself, or even a full 4-level zone. That comprehensive change forces the player to rethink their strategy for navigating the entire level. Furthermore, the weighting of different challenge types in Spelunky's different zones means that the game has a difficulty curve that allows for some breathing room. A player might feel a certain area is easier than the last, so the tension of a run isn't just a continuously growing crescendo, but has some lulls.

Zones as personal milestones

One zone might challenge the player's combat skills, another their clever use of bombs to avoid traps, and a third their platforming skills. This allows players to "rest" different parts of their brain. But on a metagame leve, it also gives them better feedback for which skills they are advancing in and which tones hey need to improve. (If you always die in the ice zone, then you have a weak spot when it comes to being patient when descending below, using enemies to break your fall, etc.) This is different from a game where all challenge types are linearly increasing throughout the run, as that makes it thus harder to tell which specific challenge types you're failing at. (In a Brogue sense: am I dying because I don't avoid strong single monsters,because I get surrounded by hordes, because I don't use telepathy/magic mapping well, because my weapons aren't strong enough, because I'm not flexible enoughi n my build, because I'm letting my allies die...?)

But really what I want to draw attention to is not to how the distinct "zones" add variety within a run, but instead, how they change the psychology of the player across runs. The structure of zones gives the player a benchmark or milestones to compare their performance between runs. If level 11 is a jungle and level 14 is a temple, then there is a noticeable difference between dying in them. That feeling is not there in a game where floors 11 and 14 are just floors of the same dungeon, maybe with increased traps and some tougher monsters thrown into the mix. The player might think "awesome, I made it all the way to the temple this time!" or "ok, I died in the temple again, I need to think about how to approach it better" at different points in their "careers".

(Nethack has something similar with its fixed/named levels - "I reached Sokoban!" - but I feel that they are far too fixed, repetitive, and spoilable. Strong theming combined with distinct procedural generation parameters is much better.)

Tying the death to a certain zone within the game's structure makes the death much more memorable, and helps the player make a mental note of where their skill level is at. It gives them an internal benchmark of their mastery, against which they're comparing not only each individual run, but also their overall growth as a player as they notice that benchmark evolve.

And that's important from a psychological perspective, because a player who notices some progress - even if minor - is more likely to stay engaged and keep challenging themselves to improve further. A player who is not improving at all - or is unable to judge whether they are improving - is more likely to quit the game. They might assume that they don't "have what it takes", or that the game has nothing more to offer to them. (This also happens with single-player games where progression is defined as getting further in the story. When a player is unable to beat a boss, they either quit, or turn the difficulty down - if the game and their ego allows it.)

How to measure personal milestones in Brogue?

Almost all games use some kind of incremental progression. In arcade games it was your increasing highscore, in MMORPGs it's your experience level, in competitive games it's your Elo rating or rank on the ladder. Note that many of those measures can fluctuate with time, but chasing the "high-water mark" is usually what keeps players coming back. So it's probably safe to say that it's a desirable feature for a game to have, there being more than just a single big milestone of success, but smaller milestones along the way as well.

That big milestone, in roguelikes, is whether you ascended with the Amulet. That's a long way away for many players. So what do we use to measure the progression along the way for non-ascended players?

• Depth level reached.
• Highscore.

Highscore is a very bad benchmark because highscore doesn't really correlate with how successfully you're playing in most roguelikes, including Brogue. It correlates with how much gold you found and bothererd to pick up, when doing so has no other risk except the time cost. Either you're picking up all the gold at virtually no cost, in which case it's nothing more than a fuzzy measure of depth level. Or, alternatively, that wasted time is significant enough to actually kill you with hunger - in which case a higher score actually means you're playing worse and are less likely to reach the amulet!

Note that Lumenstones, unlike gold, are of fixed value and spawned at fixed rates. This makes Lumenstones a more accurate measure of whether you mastered that level. Also, Lumenstones are your whole reason for being on those depths, so getting them isn't a pointless distraction in the way that gold is. Past ascension, highscore actually seems like a pretty good measure of player skill and success. The problem is that, before ascension, it isn't. For the players who die before level 26 - i.e. before they reach any of the milestones Brogue offers to them - their score might as well be a randomly rolled number with some weighting from the depth they died on.

The maximum depth level is perhaps a better measure of success than score, because it is not "fuzzed" by random rolls. But it's still not great, I would say. Maybe I feel that way because it's slightly hard to remember which depths you have reached (I play on multiple computers so I don't have access to all my game records at once). Perhaps the depths simply having names and/or stronger visual themes would be enough to help with recalling my personal best. Perhaps part of the problem is that the highscore list is sorted by the meaningless gold amount, so you can't quickly see what your good non-winning runs were. Maybe the list could also show some succinct character information, like the build or most highly enchanted item they had, dunno. Or show the icons of the allies they had. To go "ah right, that was the run where I got the unicorn..."

Ideas to consider implementing

So, some suggestions for making the game - not easier to win, but easier to find partial success in and see your growth as a low-level player. These are mostly alternatives to each other, not all to be implemented together:

1. Remove score values for non-winning runs and simply order them in the highscore list by the depth that was reached.

2. Remove gold and replace it with a fixed score for depth level reached, perhaps to be doubled if the dungeon is escaped alive.

3. Remove gold and replace them with a "mini-Lumenstone" equivalent. These are treasure with a fixed score value and are spawned at a fixed rate of one per level, for levels 2-25. The score value is perhaps doubled if the dungeon is escaped alive. This is equivalent to the above suggestion except it adds the exploration aspect to collecting the treasure.

4. Allow players to escape the dungeon at any point, even before ascension, AND give additional points for doing so. Thus, a player who reaches level 13 and knows they will likely die soon can "bank" their winnings and get a nice score, one much higher than if they died on level 14. Retreating like this is not the optimal way to play in the long run - but, inevitably, after getting several similar scores, the player will want to push themselves and venture deeper. It gives a more meaningful highscore for non-winning players to compare their best achievements with each other.

5. Add a "mini-Amulet of Yendor" equivalent to the middle depths, along with a nice mini-boss. This gives a milestone for middling players to strive for and measure themselves by, but with potentially fewer changes to scoring than the "mini-Lumenstone" alternative.

6. Allow players to trade some score for an easier game. If gold is to stay, then the obvious way to implement this would be by adding opportunities to buy equipment. Shops,vending machines, vault doors that require gold payment instead of keys, whatever.

7. Allow players to trade score for an easier game, but combined with the mini-Lumenstone idea above: the mini-Lumenstone can spawn in the cages in vaults. This allows daredevil players to skip upgrades in exchange for a higher score. Beginner players will, instead, take the upgrades since they definitely need them to survive. Hardcore players will instead have a new challenge mode:who can win with the fewest upgrades picked up. This idea can also be used with gold - simply spawn gold in the cages. It's technically the same thing,but I think that new players might assume that the gold can be used for buying things later, due to the association of gold coins as fungible resources in most games. Having a single, clearly non-fungible item, would make it more obvious that it isn't something they need to worry about.

There are others but those have to with quite significant changes to levelgen and balance so I won't mention them here.

[HomebrewHomunculus]

To also comment on what others have said:

I like the idea of selecting the difficulty mode at start; ...

In a perfect world, I would love to have multiple levels of difficulty. Today Brogue is a game that you have to loose hundreds of times before you ever win, and while purists might argue that is part of its charm (and part of the charm of roguelikes in general), it's entirely conceivable to imagine a world where you loose a bunch of times on the easiest difficulty, then finally win a few, then crank it up a notch.

Agreed and agreed. The ability to start small and then crank it up if desired was one of the use cases I had in mind for Brogue Lite. Sort of like a staircase of difficulty, where each player may find the step that suits them personally. They have some choice on whether they want to stay at a step where they're comfortable, move up to one that challenges them, or simply experience one of the steps and then move on to another game with some amount of satisfaction.

Whether the steps have the same "content" (i.e. the same dungeon but increasaed difficulty) or entirely different content (e.g. the episodes of shareware games like Doom) is orthogonal to that. Some people might not like the idea that you can get the Amulet of Yendor on easy mode - that you should be able to get something on easy, but the Amulet is the Amulet, an iconic achievement with fixed meaning. Lots of games also do the self-selecting difficulty levels with the "normal dungeon" and "postgame hell" as an optional challenge. E.g. you could beat Olmec in Spelunky and consider yourself fulfilled, and not be expected to continue on to hell and beat that as well - it's for expert players and that's ok.

Brogue already has that, but only for expert players, nothing for novice players (see my "mini-lumenstone" and "mini-amulet" examples above).

I love the idea of partial identification, and would also welcome an additional difficulty setting with full identification. Brogue Lite does that today, so apparently there is a "market" for people who would like to play in that way. For CE I'd argue that no items be removed (whereas Lite removes a bunch of things like negative scrolls, Hallucination, etc.). Why keep them? Because if "easy" modes are there to teach you how to play the "real" game, then you should see (and be able to play with) all items. Leaving them in lets you both learn how they work, learn their statistical distribution, and learn where they may or may not appear.

The problem with full identification is that some items become simply trash. Once you realize that potion of hallucination is useless, it becomes a waste of space that you have to drop each time. I'm not sure that teaching about its existence and frequency is something that's necessarily required of easy mode. In Brogue Lite, I solved this problem by simply removing the trash items. (Unfortunately, that broke the seed compatibility with CE - they now spawn different items for the same seed. I wasn't able to do a hacky workaround for this by immediately "de-spawning" the selected items when they spawn, and I'm afraid that that would have broken something anyway. This could be fixed by making a clean interface for configuring the appearance of specific items in the loot table without changing the original lists, and which would guarantee that other items' freuqencies and spawning locations are not affected. That is a refactoring that is currently beyond my skill and available time.)

Another solution to the trash items issue would be to give things like potion of hallucination some positive use - and positive enough that justifies carrying them around in place of other stuff, not just a marginal situational one.

Soft-ID or partial-ID systems could be interesting to explore, but I wonder if they would be difficult to implement side-by-side with the normal one. I also wonder if they would perhaps change the decisions involved too much - which could be undesirable, if the consensus is that decisions about item ID are a core value to be preserved.

One of the problems with item ID in general, which I think would carry on to partial-ID, is that it's a challenge of combinatorics, which requires knowledge of the set available. That is, a poker player needs to know that there are 14 cards in a suit to figure out the likelihood of getting a flush. And they also need to know the relative value of a flush to other hands. In the exact same way, a Brogue player making a use-ID decision must know: which potions can spawn, how bad the consequences of drinking a bad potion now could be, and how much value they could be losing if they drink a good potion right now. These are decisions which require familiarity with all the items in the game, and thus they inherently favour players who have played it before.

Even a partial-ID system like the one in Golden Krone Hotel, which shows the player 3 possible identities of the potion, would not necessarily solve this. A beginner player being told that there is an equal chance of a potion of X and Y type, without knowing the consequences of each, wouldn't be able to make a much more educated choice than now. Though, admittedly, the potion effects are relatively simple, so learning them wouldn't take that long, and it's easy to have a general idea from the names.

But it seems like it would make the optimal way of playing the use-ID game even more of a fixed algorithm than it already appears (to my eyes): if potion is either good or incineration, drink it on water. And so on. (I might be totally in the wrong here when I picture the optimal item ID players as simply following a flowchart that is the same for each run... but the whole reason for making Brogue Lite was, to put it in Sid Meyer terms: while I agree with (1) "item ID adds more decisions to make", I personally disagree with (2) "those decisions are interesting ones to make".)

The other suggestion, where some item kinds randomly start as identified, is probably less helpful for beginner players. In order to exclude certain possible identities for an item, they'd have to check the shift+D discovery menu - which I'm pretty sure they never do. Without keeping track of all already-identified items, they'd be as blind as now. But I suppose keeping track of already-identified items is the fun part of that system.

Apologies in advance if the way I discuss ID systems sounds sarcastic. I definitely don't mean to disregard the different ways that other people enjoy things, it's just that I have a hard time seeing the fun in this case. That's okay, of course - just like I don't see the fun in soap operas, but I still leave the people who do to enjoy them without judgment. It just probably means I wouldn't be the right person to plan a soap opera plot arc. So I think my apathy towards ID systems somewhat disqualifies me from giving any valuable input when it comes to designing an ID system... so I think I'll stop talking about them now.

This is kind of why I made Brogue Lite - I understand that many people get something out of the ID game that I don't, but I don't expect them to change their idea of fun for me. Nor want the mainline game to change to fit my idea of fun. So the fork is a way to have a playground for those ideas, without ruining anyone else's fun.

A thing I'd love to see (though I realize the implementation is challenging) is a way to alter difficulty (and play time) by "compressing" content into fewer levels.

This idea seems to be in vaguely similar area to my idea above of adding a mini-boss to the mid-depths, which can act as "the final boss" for the player's run if they choose. It would drop some "mini-amulet" which the player can escape with for some smallish highscore. Thus, playing a shorter version of the game to completion, if the player so wishes. Implementation-wise, I think adding such boss at a fixed depth would be much easier than varying the number of levels, or other such changes to the larger structure of the game.

The main thing I dislike about the miniboss idea is that it forces - or at least nudges - a player to plan their build around defeating it. At least that would be the case if it's a fixed unique monster. On the other hand, there is some set of different possible minibosses, then that's a lot more content to create and test. This is why I like the "mini-lumenstone" idea more the more I think about it: it's a goal that doesn't force you to have a certain strategy or complete it in a certain way. Just like OD&D's "XP = treasure" rule allowed for a lot more creativity and variety of approaches than "XP = monster kills" rules do. (Of course, some of these treasures could be spawned as items carried by monsters, adding some variety.)

One needs to iteratively "render" random seeds and evaluate each against some scoring criteria. The time to find a seed matching the scoring threshold is theoretically unbounded, but moreover the scoring criteria (for developers) seems difficult to communicate/audit/evaluate/refine over time. Contrast that with the challenge of writing an algorithm that begins with a given random seed and then post-modifies the level data to make the levels "easier". In practice, that algorithm would (I believe) be simpler to implement.

Agreed, but in theory it might be possible to find some constraints at the start of generation that guarantee certain properties about the dungeon without having to fully render it... but I'm not sure if that method would be any different, in practice, to the second suggestion here, the post-processing to modify the drops.

I'm picturing something like this: levelgen spawns food and strength potions in excess (compared to current levels). Then during post-processing we remove a portion of those, according to the difficulty level. This would solve my "despawning items when they spawn" problem described above. Not sure if doing it as a subtractive "add too much, then remove some" is easier than the additive "add some, then add some more". The former probably has potential for causing issues regarding vaults and other machinery that is tied to items, no? The latter additive approach could be quite simple: "where there is food, spawn a second food" and "where there is a weapon of +0 or higher, have (3-ench)/4 chance to increase its enchantment by a further +1", and so on. That sort of thing would be very safe, as it could never remove an item that a machine's working relies on?

[withinwheels]

The attraction of filtering is that it produces a 100% "normal" level that the player might otherwise have encountered on normal mode (it's just a regular seed like any other). But the downside is that the implementation is hard to develop and potentially very time consuming to execute.

We already have the seed catalog renderer, which might be modified to fill a data structure with e.g. counts of scrolls instead of printing to the screen. It can render a full seed in about a second. So no, I don't think it would be very hard to develop that aspect, but the time to sift through a few seeds may be noticeable (though it can still be parallelized.)

[tmewett]

Great discussion here, thank you all. To add a quick note: have you looked at gBrogue, which has some of these ideas, such as the ability to ascend at any time; special levels at fixed depths (which I suppose work as themed milestones); and apparently gives items a gold value when ascending?

[HomebrewHomunculus]

To add a quick note: have you looked at gBrogue, which has some of these ideas, such as the ability to ascend at any time; special levels at fixed depths (which I suppose work as themed milestones); and apparently gives items a gold value when ascending?

I haven't, actually - I think I passed by both gBrogue and unBrogue since I had the impression that they were based on outdated versions. I should probably check them out to mine for ideas!

The selling of equipment for gold sounds a bit like it's giving score for having a buff character, whereas I'd be interested in almost the opposite: having (or being allowed to) choose between getting score and buffing your character.

I'm not sure if gBrogue does something special there, or if this wiki is mistaken, but it claims that in Brogue 1.7.2, equipment also granted score.

I actually had an idea here that could perhaps achieve the goals of "up-front difficulty selection without levelgen changes" and/or "choosing between powerups and score": making keys interchangeable (usable on any door).

I'm not sure if there's some great reason for Brogue to have keys tied to specific doors, other than to prevent the player from entering the vaults in the wrong order (in which case you're still getting the same amount of items, as far as I can tell). But fungible keys would turn them into a resource, and "easy mode" could simply be "start the game with extra keys in your pocket". You could then use these to move through the levels faster, or skip some keyholders, or get two items instead of one on levels where the second vault's key is one of the items in the first vault.

If keys replace gold as the scoring item, then skipping vaults = bigger scores. Reward for risks. (Doing this would be doubly challenging if each key still takes up an inventory slot.)

By the way, is there any reason why the "item library" type vault, I mean the one with altars and falling cages, couldn't have the keys on the cages instead? Maybe I'm ignorant, but I've never found a way to get items from two of those cages at once. So, assuming the cages work perfectly, then functionally it's exactly the same outcome whether you have to spend the key on the vault door, or on the specific item's cage, right? And you open up more of the level for maneuvering. Is the cage thing just so that you can zap a staff, put it back on the altar, and then grab some other item on the way out?

[HomebrewHomunculus]

(There are also ways to do the "score for equipment" thing without simply giving points for picking up good gear - something the player does automatically during the course of the game anyway. Just requires forcing the player to go out of their way somehow to score points. Such as by picking up equipment that's clearly and visibly worse, or at least more inconvenient, than regular gear. For example: a golden sword, which is heavy and has poor damage, but gives X points, +Y for each enchantment level, at the end of the game. And maybe is protected against corrosion as a small side benefit. This is an item that's almost always a worse investment than other weapons - unless you're aiming for points. But if you are, then maybe you might as well wield it as a weapon, since you're carrying it anyway?)

[withinwheels]

So, assuming the cages work perfectly, then functionally it's exactly the same outcome whether you have to spend the key on the vault door, or on the specific item's cage, right?

The difference is, with the current design, one can return to swap the chosen item (return one thing to the library, check out another.) For example, perhaps a staff of firebolt /3 looks pretty good as an initial selection, but one depth later you find one on the floor; you might wish to back track and take your second choice.

Another common strategy is to take a weak weapon or armor in the hope that it is runic; after using it for a while with no apparent abilities it can be returned for a more known quantity like a wand or charm.

[withinwheels]

Another fun but minor quirk of libraries is being able to use charges of useful wands (say, empowerment or plenty on allies) while in the vault with the full intention of then turning them back in and making your real selection.

[HomebrewHomunculus]

Thanks, withinwheels. I suppose the "borrow the item and bring it back" aspect is quite integral to them, which I probably should've realized from the name "library" :) I never used that possibility much myself, since I find backtracking a bit tedious. Very good point about the wear-ID'ing runics strategy though - that one actually seems a bit more interesting to me than most ID strategies.

On the whole, though, based on your answers, I think I could quite happily change it to "keys used on item altar cages" in my fork without degrading the experience in a significant way. (But, as stated, that's only from the perspective of someone who plays in a quite naive tactics-focused way and doesn't tend to use these strategies anyway, so I'm not a very representative sample.) Thanks!

[rjsands]

What if something was done similar to Duskers, Invisible Inc., etc., in which there are settings that can be adjusted to tailor the difficulty to each player's preference? Maybe that's a separate task from redesigning Easy mode itself. I have been playing around with having potions and scrolls auto-id, but leaving staves, wands, and weapons unidentified. Personally, I find these games much more enjoyable than the standard game - like @tmewett said, I'm not making the same use-ID decisions, but I feel like I can make more tactical decisions about the current situation and what I have available. I like the idea of being able to quickly change settings for different games depending on the level of challenge I am feeling up to.

I haven't had a chance to play with much else outside of auto-ID of some item categories, but maybe there could be similar settings for number of monsters on a level or their hit points?

EDIT: For example, add a Settings option to the main menu, and selecting that might give: Auto ID:

1. Potions [x]
2. Scrolls [x]
3. Rings [-]
4. Staves [-]
5. Wands [-]
6. Weapons [-] Other options...?

[DevJac]

This is 3 years old but still relevant.

I propose that the only difference in Easy Mode should be extra items.

Players would have more consumables to experiment with, higher enchantment levels, more health, and more strength.

Let's talk about pros and cons. I'll give the cons first, because I don't think there are many:

Cons:

1. Player will get used to having a using lots of items. (This isn't necessarily a con though. See the first pro.)

Pros:

1. Players will get used to using lots of items, and when they switch to normal mode they will be accustomed to using items, which will help them avoid one of the most common roguelike mistakes, not using consumables early and often.
2. The difference between easy and normal mode might only be 1 number in the code. The number of items, or probability of items per floor. This might be extremely easy to implement.
3. The game will be easier, but still fun and challenging. In fact, some players, who prefer to win more often and without as much challenge might play easy mode forever. These players will still get the full Brogue experience. Still identify items, still worry about that first Ogre and Wraith, etc. They just wont have to be quite as optimized in their build.

If I can get some assurance that this would be accepted, I will look into implementing this myself.