Alternately, you could tell the player the exact location of the treasure, even mark it on their map, but leave a sleeping dragon/ogre/whatever on top of the treasure.
Player still has to travel to the treasure location, but taking the treasure is no longer "free", as they have to risk combat/use pickpocket/etc.

that's what it does now, but with only a 50% chance of being guarded. the idea being that guarded treasure is still there because its guarded, and unguarded treasure is still there because you're the first to ever find it (somewhat contrived).

if its always guarded, its no longer a MOVETO_QUEST (move to rumored treasure location), and becomes more or less identical to the ANIMAL_WITH_TREASURE_QUEST.

with the MOVETO quest, i was looking for something the player would want to MOVETO as a quest. treasure seemed the obvious choice. also, some caveman rewarding you for simply going to a location for no good reason made no sense. thus the rumor of great treasure was born - IE a reason for the player to WANT to go there. i added the 50% chance of it being defended to keep the player on their toes, as they'd never know which rumored treasures they went after might be guarded. its a quest generator, so they can get this type of quest again and again and again, along with the other 49 or so types of quests. and each time, it will be a different treasure in a different location, and possibly guarded by a different type of "monster". the quest encounter tables are terrain specific, there's about half a dozen tables as i recall, each with one to two dozen types of "monsters". quest locations, treasures, and monster types are all randomly generated.

and quest number appearing, quest treasure quantity, and quest treasure quality all scale to player level (party strength actually).

you see, i finally figured out what to do once you're high level - its not "what to do once.you're high level", its "whats left to do once you've done it all" - IE when you've run out of content!

the player needs a continual source of compelling new content appropriate for their level. in tabletop classic D&D, not that hard, given an obliging DM. in a computer game - a wee bit tougher. <g>.

the answer of course is the usual stuff, procedurally generated content, regular releases of new DLC, mods, user created maps / missions etc.

so for caveman the plan is questgens. i came up with ideas for about 50 so far. version 1.x had 16. i've already done 20, with this MOVETO quest being the only one thats not really ready to go yet. besides questgens, a simple quest scripting language and quest runner (mini VM) have been designed, allowing both user created and DLC quests. i plan to add quests as an encounter type that can happen with any caveman, not just at a band's shelter. hopefully the combo of lots of questgens, a game based on random encounters and random events (instead of hard coded spawn points and NPCs), and quest challenges and rewards that scale with player's level will help keep things fresh for long term high level play.

so some, if not many of the treasures would be empty b/c someone else had already heard the story and gotten there first.

ooh, i like that idea!

looks like the MOVETO quest just got a new random optional plot twist!

when the quest is generated, one of the 20 generic int variables available to the quest would get used as a TREASURE_ALREADY_TAKEN flag, which is set at random when the quest is generated. if the flag is set, when the player arrives at the location, no treasure or monster spawns. instead they get a message like "no treasure here! looks like someone beat me to it." and the quest ends.

i was working on a branching version of ANIMAL_WITH_TREASURE that included similar twists such as you weren't the only one going after the treasure, but you were ahead of / behind / encountered your competition on the way (or the way back). it included the treasure was already gone, and nobody got anything. it got really complicated really fast with all the things that could happen. it was about a 5 stage quest that usually branched three ways at any stage, so a 5 level tree with about 3 to the 5th = 243 leaf nodes. eventually i decide to start with simpler quests.

Give clues to talk to a few people, and let them tell half-stories that the player has to combine and make sense of.

hmm... difficult to generate. puzzles, riddles, and mysteries always are.

i'd pefer to stay away from the whole "story generator" thing and save more storyline based quests for the quest scripting language and the quest runner mini VM. scripted quests will be able to call functions to generate random locations, monster types, and treasures, but will otherwise be pretty much hard coded. this makes them more suited to heavily storyline oriented quests. storyline quests are like a book or a movie, a hardcoded one shot disposable form of entertainment. they can be generated, but its a fair amount of work, especially to get the writing sounding good. and generating mysteries and riddles - even harder.

it could be something like, "Tracks here show that was here, and head northwards", or "Small bits of are scattered here, you must be nearing the treasure!"

a trail of bread crumbs... often makes for a fun quest.

I'm thinking maybe something akin to the old 'getting warmer/getting colder'

i like the idea, but you'd need a series of landmarks (or something) as your bread crumbs. and that gets into creating a bunch of objects for them to stumble across.

auto trigger when in range is a possibility. that's how quests start. you don't go up to a caveman at a shelter and say "hey dude! got any quests?" <g>.

if your within 50 feet of a shelter a quest encounter can occur: "the cavemen here tell you about <something>. go after it? yes/no"

when you get within 100 feet of a quest location, you get something like "you found the animals rumored to guard a great treasure!" and animals and treasure spawn.

when you return to within 50 feet of a questor's shelter, you get a message like "they thank you for your help and give you a reward".

so you never actually talk to anyone. you don't have to select an NPC and do dialog. everything triggers based on range to quest origin or quest target (usually). one really nice aspect of this is that the two standard "talk to band" or "talk to caveman" dialog menus don't have to be able to handle variable quest dialog. keeps the code simpler.

so maybe as range to target decreases, it could trigger messages indicating you're getting closer. and quest variables could flag which messages had been triggered. or maybe track the player's closest range to target in a single quest varaible, and trigger messages as that passes closer milestones, and show a 'getting cold" message if range goes far above the closest range yet.

but to tell you the truth, i'm not entirely happy with range triggered quest messages and dialog menus. its an immersion thing. you should have to talk to someone, or they should talk to you. range based was just a simple way to handle it generically. but they may be ok. reminds me of the stolen painting mystery quest in bruma in oblivion. they did a bunch of breadcrumb items in combo with range based messages about what you were thinking to yourself.

or something like the old 'Guess Who?' game, where each hint is something like, "The treasure must be on a mountain top", "The Treasure is on a mountain with no trees", that help narrow down which specific location the treasure is on.

that's a puzzle / riddle. difficult to generate. also has zero replay-ability once solved, just like a hard coded puzzle in a game, or a hard coded riddle in a storyline quest. so if that randomly generated combo of clues appears again in a subsequent quest, the player already knows the answer, ruining the quest.

gotta remember its a quest gen, not a single hard coded quest. MOVETO treasure is the type of quests it generates. in OO terms its a class, not an object, and definitely not a singleton! <g>.

granted, more clues and more random choices for each clue makes for more combos that will take longer on average to repeat. but more non-replayable combos doesn't make it replayable, it just makes it takes longer to use up the limited combos of one shot disposable hard coded content it can generate.

Though a craftable dowsing rod might also be an option unless it's too magical for your game.

dowsing is not magic. science can't quite explain it yet, but it appears to be a real physical phenomena - perhaps related to ESP.

so yes dowsing is a viable option - but you still have to search... OTHO, a dowsing rod is like a compass... could make searching a whole lot easier, huh?

but dowsing is also a "gift", like ESP. so the average PC would have zero skills... ugh!

sometimes i hate working on this game! realism makes everything so HARD! things get complicated so quickly! sometimes i wish i was making a game where i could just make up the rules, and as long as it was cool and fun it was ok. [its a] whole lot easier than this writing [of] simulator stuff [is].

For the first landmark tall mountain or large waterfall would do, then you need to go around it until it has distinct look (a rock that looks like tiger if you stand directly north to it) Then you look for fallen tree that bla bla bla and it's buried 10 steps west from it.

clues and breadcrumbs, again difficult to generate. breadcrumbs or a trail of landmarks isn't too hard to generate - not as hard as clues at least.

but in the end, one landmark, a trail of landmarks, dowsing, or a simple x on the player's map are all just different ways to get them to the location. some more immersive than others.

once there, should they have to search? should it just be there? perhaps guarded, perhaps not? perhaps a limited search? (spawn treasure somewhere nearby).

all the other quest gens use X on the player's maps to indicate location, and no compass indicator (cause there's no compass! <g>), so i doubt i'll generate landmarks just for this one quest gen.

while "x marks the spot" and range triggered quest dialogs are not the most immersive way to do things, its the gameplay of the quest itself that's the real concern. maps and dialogs are just a way to give necessary quest info to the player. they don't (or shouldn't) really affect how the quest plays out.

There's also what I'll call the "Assassins Creed" approach, whereby you just put a big circle on the map and say "it's somewhere in there, run around till you find it".

Not one of my favourites, but it gets the job done.

B.G. quite true!

its a quest gen, so i'd need something like a list of a dozen or so types of landmarks to choose from at random... and they'd have to be unique somehow - just another tree model in a woods map square that has 1300 trees in a 800x800 foot area woudn't be much of a landmark.

for the GET_SPECIAL_ITEM questgen, i have a list of ten "special items" to chose from. but they are just a quest variable, IE you have it or you don't, and its type (and thus its name via lookup table for quest dialog text). they're not a world object or inventory item. you can't equip them ir interact with them. just get them, and give them to questors.

but regardless of how the location is indicated, via a physical landmark in the game world, or a mark on the player's world and local maps, or both, once you arrive at the location, there's still the searching to be done.

Hmmm.... that does sound like quite a bit of work. Setting up who knows how many landmarks for your questgen to choose from, unless there's And further, the danger of them all being the same, as though its the only landmark cavemen ever use. I don't know if this is what you just said, but you could use a map marker to mark the landmark ('This is the tree Tiger Killer told me about.') and then have the player search around that landmark for the spot to stand on, which would generated some range of feet away from the landmark. Also more work, as you'd have to mark your potential landmarks as coordinates, but more in the range of the do-able. I have no idea if marking the trees with claw marks or unique/rare fruit would be worth the effort. I'm beginning to suspect not, that you have a quick and clean design for your randomized generator, and that makes sense, you want to keep that.

it really not about how the location is indicated to the player. i have no real problem with x marks the spot

its about what they should have to do once they get there.

some possibilities:

1. they don't have to do anything. the treasure is there, if unguarded the quest ends. if guarded, quest ends if all guards killed. if player moves far from guards and they are removed from the simulation, they respawn when the player returns to the quest location. this is the same as ANIMAL_WITH_TREASURE, except there's a 50% chance that there is no animal, just treasure.

2. treasure spawns near the quest location, but not within sight of it. player must wander around and find the treasure. if guarded, guards spawn/respawn when player approaches. quest ends when treasure found (unguarded) or guards dead (guarded). this adds some basic searching to version #1.

3. treasure spawns not very close to quest location. payer must perform organized long term search (IE search pattern). version #1 with lots of searching.

4. player doesn't have to search. simply being near the location assumes you're searching. after a random amount of time, treasure and any guards spawn and you get a message "you found the treasure". this is how it works now. not very realistic.

5. follow a series of clues to find the final location of the treasure (cue the Riley quote! <g>). clues can be hard to generate.

6. follow a trail of landmark breadcrumbs to find the final location of the treasure. can be a lot of work to implement for a generic quest generator. need lots of combos of landmarks to reduce repetitiveness.

approach #2 doesn't sound too bad. more or less the same idea mentioned by swiftcoder. does add some searching, without requiring 44 hours to search a whole map square.

maybe place it at random within a 1000 foot BBox radius. but then it might spawn withing sight. place it at 700 to 2000 foot range (at random) and in a random direction from the quest location. max visual range is 600, so it would always spawn out of sight, even in open flat terrain.