All content on site copyright © 2021 Caravel Games, All Rights Reserved, unless otherwise indicated.
Construction Methods (by Roger Barnett)
While I'm no expert at creating holds, I'm no amateur either. So I thought I'd write a little guide to the way that I produce my rooms. These are my own construction methods, not any official ones and contain my opinions rather than those of the forum (though they may concur here and there ). Hopefully this little article will be of some use.
If there are any mistakes or you would like to contribute anything please contact me.
- Lynchpin based
----- Multiple Goal
- Action based
----- Remote Manipulation
----- Movement Order
- Cheap, Unpopular and Filler
- Spicing up the basic rooms
- The Assistance of Others
- Surface Polish
These have been superbly covered by Erik in the helpfiles of The City Beneath so I won't reiterate them. Importantly - make the rooms fun. Fun is a lot better than just plain difficult.
Make sure that all rooms are accessible, it's not funny when there's a room you can't get to at all, even if it's optional. Don't make cheap rooms. Don't use 500 monsters where 50 will do. Make sure that if the room is not solvable from all entrances that you can manuever through the room so you can enter from a winnable entrance.
Mistakes to avoid - make sure there's no required rooms on the other side of the blue door - you need to be able to get to the exit. If you have an area that's only accessible after dropping the Master Wall, make sure there's no rooms marked Secret in it. Make sure there's no places for the player to get stuck - areas that he cannot escape from without restoring to an earlier room - this can happen with badly designed clone puzzles.
It's hard to have too many checkpoints in a room.
Make any storyline scripting/NPCs unobtrusive, the game is about the puzzles more than the story.
One other more subtle thing that can catch people out. It might happen that after you've made a room, you want it to fit into a level you've already created. And so, you reflect it vertically or horizontally. Make sure to test it again after the reflection, just because it works one way round doesn't mean it'll work the other. Yes, the majority of elements in DROD are symmetrical and behave the same no matter what way round they are, but there are a few things in which there is a preference for certain directions. Goblin movement is one example, a far more prevalent one is snake movement. I've had a brilliant puzzle involving an adder that only worked if the room remained in its orientation, changing over the horizontal/vertical completely screwed it up.
The same goes for decorations. Adding decorations to a room (obstacles instead of walls, pits instead of water etc) makes it a lot nicer (see the section on Surface Polish) but it can screw up the room - test it again once you've decorated it. Two of the biggest effects are briar on water/pits, and seep on walls that have had obstacles added atop them.
And finally, no matter what you do, you're going to make some unfun or unoriginal rooms. That's not a problem. Just remember to make more rooms than you're actually going to include in your hold, and trim out the worst offenders. You don't have to include every room you've made. Perhaps keep a master hold where you create your rooms, and piece together a hold for submission out of the rooms you make in there.
Logical and straightforward. This method is probably one of the harder ones and takes time to get right, but it does tend to lead to some of the best rooms. The method works in reverse - firstly you design the last stage of the room, adding layers of complexity as you go backwards.
Let's take an example - a room from the classic Architect's Edition hold - "Simon's Dungeon"
Simon's Dungeon : Second Level - Death on Wings : 2 West
This is an excellent specimen of a fun linear room. The goal is at 3, to kill the wraithwing behind the door. You would thus design something like 3 first. There are several possible endings. Having a snake in a loop that needs a monster in it to kill the snake. A mimic potion that allows you to trigger a distant orb. A tarstuff mother. The last trapdoor attached to a bridge that makes it fall. Anything that once done virtually completes the room (there may be some clearing up of monsters after)
So, you've got an endgoal. Work backwards. In the case of the above room, the next step of construction is that you can't get to the wraithwing. You need a second wraithwing to encourage them to get close to you.
Fine, you've got a second wraithwing. But you want to add a further layer of complication. Put the wraithwing at 1, where you have to perform some tricky long-distance manipulation to get it out of the area.
You've now got a second wraithwing to solve the puzzle. But you want another layer of complexity. So you add in something like 2. You have to manipulate the wraithwing through this in order to get it to the goal.
This is the core idea that's useful in the method of linear construction - build a goal, then work backwards, adding complications to get to the goal. Once you've gotten a few layers of complexity you can always work forward from the start, adding an initial puzzle. It's one of the harder methods of building rooms but it usually leads to good ones. Just remember that just because the sequence of events that you have to carry out to complete the room is obvious to you, it might not be to the player.
Though it sounds similar to the linear method, this one produces quite different rooms. Once again, we'll use Simon's Dungeon as an example - this room in particular
Simon's Dungeon : First Level - Roach Slayer : 2 South, 2 East
The decision tree is flatter in this one, and the room is a little easier. Rather than having a linear series of goals, you have several that must be done. In each case - leading the roaches into a trap. Once that's done you can open the doors and have yourself a little killing spree. The key difference between this one and the linear method is that it doesn't matter so much what order you do the goals in as long as they get done.
To design a room like this, you simply add in lots of things that must be done, and don't force as much of an order onto the player. I say 'as much of an order' because the two methods aren't absolute - it doesn't have to be either linear or flat. You can blend the two - say you have four orbs that must be triggered. Three of the four can be triggered in any order relative, but the fourth must be triggered after the second or a roach walks into an arrow trap. Stuff like that.
If you're short of ideas for goals, then the DROD Idea Generator is handy.
You don't plan to make this kind of room, it comes upon you randomly. It's based a lot on having a sudden idea. The idea behind is that you take two or more elements that aren't usually used together. And then you use them together, exploiting their interaction. This can then segue into any of the other room types, but the core of the room is based around this unusual interaction.
e.g. Rock giants and adders.
The interesting quirk here is that although adders can eat the rock golems, they cannot eat the rock giants. This leads to all sorts of interesting puzzle ideas. For example, have a small adder in a crowd of rock giants - you have to find a way to get the adder out somehow. Or have a space that needs an adder of a certain size, you have to feed it golems and golem remains, but you can only kill one or two rock giants first. And they must be the right ones.
This sort of room is hard to plan to make - it will just come upon you when you think of an unusual interaction between elements. Quite often when you come up with the puzzle, you won't need to add that many added layers of complexity. The puzzle works well on its own.
This is a more extreme version of the Quirk method. When making this type of room - give yourself restrictions, rather than letting your creativity flow free.
For example - make the room small and try to make the best room you can in that very limited space.
Pick a single monster type and only work with that monster
Restrict yourself to only a single element and work with that.
Theme the room around a single element.
This method is more tricky and isn't so good for novice architects, I admit. But a good architect finds the limitations a challenge rather than a restriction. It also helps sometimes, since you've not got as many paths to go down, you can focus more clearly on the path in front of you. Take a look at the Limited Space competitions on the Contests board for examples.
A type of room that I'd not included in here, suggested by vinheim. It can be considered a separate room type. The idea behind is this: Setting stuff up in advance.
It could involve dropping certain doors before you enter a room, so that a brained roach can make its way to you. This is probably one of the simpler examples, and has been around since KDD. But, there are a lot of possibilities. Making sure you leave a trapdoor path open that you can cut through later. Directing a a creature to a certain point before you go inside.
But, the entire idea of this one is that you do most of the work before you 'attempt' the room. And that you cannot directly observe the effects of what you've done until you go inside. JtRH did this with bombs and doors to alter the paths of the fuses. You could also set up decoys before you enter a room. The possibilities are many - but the entire philosophy behind this kind of room is that you set stuff up before you do the main core of the room. If you've set it up right, the room works out easily. If not, the room is unsolvable.
Be careful with this kind of room - make sure that there's some kind of logic to the room. If it's just trial and error, hacking around with every single different option to see what works... it's not going to be fun. You need to be able to logically deduce what changes you make on the outside before you go in.
This is a somewhat... cheap method of room construction. It's probably the quickest method even though it's iterative, but it also leads to a very great variation in the quality of rooms produced.
Throw down some walls, obstacles and other stuff (orthosquares, force arrows, tunnels etc). Bombs that set off here and there. Maybe a spinning mimic or two. Then add in the monsters. Tar mothers especially, roach queens, water skippers scattered all over the place. Brains are optional, it depends on the room layout. Make it so that there are lots of avenues by which you can get to the monsters and the monsters can get to you. Then test the room out, see how hard it is, whether or not it's possible. Then make some alterations, perhaps fix a roach queen in place with a force arrow, add in a tarstuff switcher token to make cutting different. Then try it out again. And again until you've got it. A speed position or a mimic can make handling the swarm a different experience. One thing that's often tried is monster generators (roach queens, blobs of tar) behind red doors, inaccessible till you've dropped trapdoors all over the room, but will still send monsters in your direction.
This method requires caution. It can lead to some really really cool rooms (Pirate Hideout in The City Beneath, for example - horde management with speed potions), but it can also lead to some really boring hack and slash rooms. The best thing to do with this method is to keep the battlefield rooms separate. Make up a few of them. Try them out, get someone to beta test them. Then the most important bit - throw out the bad ones. If you make six rooms using this method the odds are that only one or two of them will be good. The others will be alright at best.
A somewhat random method, but I've used it with great success to make some really cool rooms.
Slap some walls down randomly, don't plan.
Slap some movement restricters down randomly (force arrows, orthosquares, tunnels, trapdoors), but less of them.
Slap some monsters down randomly.
Put something else unrelated down - perhaps tar, or bridges or even citizens to get in the way.
Then test the room out and see what happens. If nothing cool looks like it's happening, delete the lot and start again. If you see the potential for something, or something cool happening (monsters bunching up at a point, things being triggered, spots where you/the monsters are safe) then slap some more random stuff down but avoid the areas that looked cool. Test it again. See if there's any cool stuff happening in the new areas. If so, keep them, if not, delete. Repeat until you've filled at least half the room.
Then tidy up the random stuff, make the room look smoother and test it out.
This method isn't quite as cheap as the Battlefield method, but it does tend to produce interesting rooms more. It can be very frustrating at times, especially if nothing interesting is appearing. So only do this method if you're bored. Once you tidy the room up though, no-one can tell it was a scattershot room.
An often used method that produces some good results. For this method, forget about doing much to the monsters directly. The point is to move them around so they do stuff themselves. There's a few subclasses of this one
1) The Maze - draw up a random mazelike structure with only a single exit, or a hot tile in it. Put a monster or two in it, and see if you can make them escape and how easy it is. This one can be tinkered with a lot, with doors, pressure plates etc. Or by having multiple monsters and needing to have them interact with each other. e.g. a roach queen whose path must be stopped for a single turn by one of her child roaches. Put in a few arrow traps or otherwise to make the problem more difficult.
This has a lot of variety, since there's several monster movement patterns you can use. A roach is the standard, but having a wraithwing in the maze means you can control it slightly from the outside, same for a goblin. Guards are interesting too, since they rotate before moving, allowing you to buy a turn or two. Same for aumtlich. Slayers are also interesting in this regard, since they rotate their hook upon hearing a tunnel noise. Wubbas/golems/gel are more restricted in their moves too.
King Dugan's Dungeon : Nineteenth Level : 2 East is a good example.
2) The Puppet - take an area with a monster or monsters in. Then have them do things that affect the outside world. e.g. have a guard in there who must kill all the monsters. Have a slayer who must do the same. Have pressure plates that affect doors on the outside. Have orbs that a guard/slayer must strike. Have an adder eat the monsters. Open doors to guide a stalwart around. Move bridges to alter the path of snakes. The possibilities are nearly endless.
3) The Mimic - an old classic and simple to describe. Have an area with one or more mimics in and use them to kill monsters or set off orbs/trapdoors/pressure plates. Works well when combined with the Linear/Multiple Goal methods.
King Dugan's Dungeon : Twenty-Second Level : 1 West is a classic example of this one.
4) The Clones - A clone allows you to direct monsters. This one doesn't have to be in complete isolation - there can be monsters all around, but the clone or two allows you to move them away from one Beethro. Don't go overboard with the number of clones though, or things tend to get quite boring.
5) Door manipulation - Use this to move monsters around - works very well when combined with Brains/guards as the monsters can pathfind to you.
Don't be afraid to combine elements of the above into a remote manipulation puzzle.
This type of puzzle is hard to categorise. However, it relies on the order in which the monsters move. These used to be not much in favour back in the days of Architect's Edition, but they are coming back into fashion.
To begin with, remember that Beethro always moves first. Before anything else. His mimics come next, and then the monsters. The monsters will move in the order they have been placed in the room in the editor. For example, if you place down three wraithwings and a gel baby, the gel baby will always move after the wraithwings.
Simplicity is the key here. Requiring the player to right click on fifteen different monsters to see which moves first is not fun, and shouldn't be done. Keep the movement order sections to two or three at the most.
There are two simple uses of movement order that I've found to be very useful. One is to hold back the "non-diagonal" movers for a turn in some way. These are the wubbas, the gel babies and the golems. It is possible to engineer it such that they remain stationary for a turn if they move first. They will bump against the other monster, which will then move. Yes, this works for creatures like roaches but the effect is usually more pronounced and easier to engineer if the creatures cannot move diagonally.
The other movement order puzzle usually involves serpents shrinking by means of a temporary dead end. The dead end only lasts a turn before the monsters move, but it's enough. I've used this one myself.
This puzzle works well with the Scattershot or Battlefield approach to puzzles. Often it's not something you plan, it's just an effect that shows up which you may exploit later on.
A warning though - don't try to set up movement order puzzles in rooms which already contain a lot of monsters. If you make a mistake, deleting and re-entering a load of monsters in the right order in the editor is an absolute nightmare.
This one deserves a subclass of its own. It is entirely possible to make unusual or scripted creations that aren't part of the normal DROD rules. A good example of this are the Torture Chambers in TCB, with spinning and mobile mimics.
The most important thing about this class is that the scripting MUST BE OBVIOUS. It's very bad form to have hidden stuff going on behind the scenes. Any moving NPCs that you have to interact with directly (stab, touch, whatever) should have consistent logical processes. Have a mimic that takes two steps forward and two steps back repeatedly - fine. That's logical and you can predict it. But don't have a mimic that journeys to random spots on the board, you can't predict it.
If you're going to make a customised monster, either explain the movement pattern to the player or make it obvious. This can lead to some very interesting puzzles - NPC monsters are not affected by invisibility potions, decoys or brains. Furthermore, if you're going to have an NPC monster either design your own sprite for it, or use one of the 'human' sprites - if the NPC is not distinct visually from the normal monsters your players will be unhappy.
If you're in doubt - simpler is better when it comes to scripting. Complex scripting often has a way of going wrong. Make sure that you carefully include the "End" commands if you don't want the scripting to be always there, otherwise you're going to have breaks. Even something very very simple can lead to interesting puzzles - e.g. a monster that moves up and down every 5 turns. Put it in a tunnel with an orthosquare above it, and you have a gate that's open for 5 turns, closed for 5 turns, but can be held open by having something in it.
Cheap, Unpopular and Filler rooms
There are some rooms that have fallen out of favour and are regarded as really not being that fun. There are obvious exceptions to the rule, but when making these types of room take care and accept criticism on them. It can be summed up in a single sentence - "Don't make the player hate you."
1) Hidden things under tarstuff - If you've not included a vision token, this room type can be remarkably obnoxious. Trying to find a pressure plate or a trapdoor under tarstuff is not fun at all, and will quickly annoy your players. King Dugan's Dungeon : Eleventh Level : Entrance - a good example of what can be annoying - there are a load of walls under that tar. Vision tokens didn't exist when this room was made, and it annoyed people.
2) Orb/Pressureplate Door Puzzles - These work well if they're not too complicated. If there's too many possible combinations then the player will quickly get annoyed. I've made this mistake myself.
3) Rooms full of hordes of monsters with no challenge - These are annoying. Just clearing out the room isn't hard at all, it's just dull. Don't do these.
4) Rooms that are far too easy and contain monsters just for the sake of it - Again, don't do these. If you want to fill out your level graphically then include linker rooms without monsters. Don't include monsters just for the sake of it - you can include empty rooms
5) Large amounts of tarstuff to wade through - this can get really dull really fast. If you're going to include this type of room, make sure to include something else with it. Say, pushing a mirror through the tar. Having to switch between tarstuff types to make progress. Something. But rooms like King Dugan's Dungeon : Eighth Level : 7 South, 3 East have really fallen out of favour.
6) Large Mazes: Think Level 13 of King Dugan's Dungeon. Not fun at all. If you're going to do a maze, try to limit it to a room or two at the most, or make it a secret level. Tunnels and doors mean that you can have excellent complexity and depth in a single room maze, you don't need it sprawling across a level. I freely admit that I've broken this rule myself, but in my defence it was a secret level and you were warned repeatedly that the level was evil, optional and you shouldn't go in
7) Rooms with a lengthy setup where you have no idea if you've gotten it right until near the end. e.g. say you have to arrange various trapdoors and doors around yourself, and then trigger a massive brained roach hoarde. You won't really know whether you've got it right or not until you trigger it, and even then you still need to test that you've definitely got the wrong solution. Trial and error rooms are not fun.
8) Darkness hiding things. Light and dark squares are there to add atmosphere. They are not there to hide things and make the puzzle harder. Exceptions can be made if you leave certain notes but on the whole - avoid screwing around with light levels as part of a puzzle. e.g. monsters/pressure plates hidden by darkness are a bad move.
9) If the room isn't fun, don't include it. Simple as that.
Spicing up the Rooms
So, you've made a room or two using the above method? Good. But it's a bit boring? That's a shame. Add in some complexities to make it more interesting. Don't rely on these ones though, these are not intended to be the core of a room, just a little spice on top.
1) Orthosquares - these add in a nice bit of mobility restriction in certain areas, making them more difficult. They can also be used to delay monsters a little.
2) Turning traps. Build one of these by having the player start with a mimic potion. The mimic must be placeed on a pressure plate surrounded by bombs on many sides. Have these bombs make the room impossible if triggered. This forces the player to keep his sword in a restricted orientation, making the room harder
3) Timers - roach queens in isolated areas that gradually fill unless you can get a mimic in. Tarstuff that grows and spawns an impossible-to-get baby. Briar. Snake twists. Be careful with these - in many rooms all a timer will do is make it more difficult, not more fun.
4) Hasslers - monster generators that are inaccessible until you've solved the core of the room. This makes you step a little more carefully as there's always a steady supply of monsters coming your way
5) Trapdoors - If when testing the room, you find you only need to cover a patch of ground once or twice... add in some trapdoors there to make it more restricted. For the player who doesn't know the logic, this makes it more interesting
6) Crumbly walls (not secret ones, crumbly ones) that can be broken to allow new monster paths and allow you more mobility
7) A Slayer or two. Works well in battlefield rooms as not only do you have to fight the hordes but avoid a Slayer as well. Of course you can always use him to your advantage.
8) Hot tiles when you're fighting tar babies. Forces you to keep moving
9) Aumtlich placed in initially inaccessible places to provide a hassle from outside the arena. Same for evil eyes that must not be woken until later.
10) Pressure plates that must not be stepped on, or the room becomes impossible. Or the converse - pressure plates that must be occupied while you sort something out.
11) Stalwarts that must be kept alive. They're dumb, so be careful with them
12) The occasional patch of oremites to allow a quick direction change - very interesting when combined with 2.
13) Tarstuff that must be trimmed back occasionally or the room becomes impossible. The mother should be elsewhere.
14) A wubba or two to get in the way. Works well in narrow rooms.
The Assistance of Others
So... you've finished your hold. It's got a few levels, each with a few rooms and you think you've finished. Oh no you haven't. This stage will vary depending on whether you've been posting gradual upgrades on the Architecture board (or in a private beta) or whether you've completed the entire hold before showing it to anyone else.
Case 1: You've been posting it in stages.
At some point... you do have to stop. You have to cut the hold off and say "That's enough, the hold is finished". Holds can drag on way too long and be unpleasant as a result - a long hold isn't necessarily the best. e.g. The Choice - very short hold, and yet worthy of being a Smitemaster's Selection. Don't let it go on too long, there are far too many unfinished holds on the Architecture board - don't be one of them! Stop, and get it fully betaed. See the Ambition section.
Case 2: You've posted it as a completed work
Almost everything here will apply to Case 1 as well. Now, as obvious as it might seem, I'm going to repeat it. Listen to what your beta testers say. Listen to what your beta testers say. Listen to what your beta testers say. Now, obviously you're going to have to correct things like backtracking issues, spelling mistakes, unintended solutions and suchlike, but frequently if your betas are doing the job properly, they will comment on the rooms and levels. Listen to their comments and try to respond to them.
General rule - if a beta says a room is boring, too hard, or just unpleasant, take another look at it. Tastes vary, so it might just be them - if you really like it then, keep it. However, if two betas agree that a room is bad, you're better off removing it. It's the rule I use myself, if two betas say one of my rooms is bad, no matter my own personal feelings on the matter, I either remove the room or retool it according to their suggestions. Listen to what your beta testers say.
Don't be afraid to cut out rooms. Save them elsewhere to be worked on, but unless you're some kind of architectural god, not every room you make will be genius. Expect to cut out several rooms from your completed hold if you want it to be great, you really don't have to use them all. Better a great medium-sized hold than a mediocre large hold.
Patience is another tool - not all of your betas will progress at the same speed. Some might plough through the hold and give you comments in three days. Some might take three weeks to get through it. You must unfortunately accept the fact that on a private beta board, not everyone who agrees to the beta-ing will actually play through and comment on your hold. Some might only do the first level, some might not do that. Provided you get one or two decent betas though, you should be fine. Also, quite a few of the names on the "willing to beta" list haven't logged into the forum for a while - if they don't accept the beta, don't be offended.
Once at least 2 people have fully played through the hold, and all the mistakes they've found have been corrected, you should be OK to submit it to the HA's. Don't expect a beta service from them - their job is to check that the hold is completeable with no backtracking issues or game-stopping problems. If they spot them, they'll probably point out unintended solutions - but it is not their job to spot them. The hold is supposed to be tested and betaed as much as it can be *before* it goes to them. They're a busy bunch.
This part should be done last, when you've done the room and are happy with it. There are six main graphical themes in TCB, and while others can be made, you cannot rely on the end player having those themes. Feel free to improvise with image floors/walls.
Give each level a theme and stick with it. Outside? Inside? Caverns? Castle? Dungeon? Once you've done that, decoration is easier. Just changing the floor type can make a level look totally different - two levels can both have the Deep Spaces theme for example, but look distinct. One with a rough dirt floor and rocks everywhere, the other with a snazzy mosaic floor and regular statues. It sometimes helps to design a master layout of walls and pits to make the level map look cool then fill in the puzzles after. Take the first level of "The Consequence" as an example.
In many cases (but not all), pits, water and obstacles are functionally identical to walls, so you don't need to just use walls to bar the way. Add in some totally extraneous walls/obstacles to decorate the place. Light sources work well to hilight certain areas.
Write a good blurb for the start of each level. Even if you're including no dialogue in the hold, you can write a paragraph to describe each level. It really helps.
There are lots of talented voice actors on the forum - if you want to include a storyline, all you need to do is ask and you can get several voices. Same for graphics - if you want a custom sprite/avatar, just ask and there'll be someone who can do it. Better yet, do the voice/sprite yourself if you can, it'll add uniqueness.
Remember that surface polish will make a good hold better, a great hold greater, but it will not rescue a boring hold from being boring. Polish is good, but add it after you've done your best with the puzzles.
So... you've decided to make a DROD hold, eh? Good for you. I hope this guide is of some use. Now, onto ambition... and the size of your work. There's two points to consider:
Firstly... if you're newish to hold building, do NOT construct more than one hold at the same time. Just don't. It won't work. Yes, you may have a great story in mind for one, and a very cool idea for the other. But... in practice what often tends to happen is two or three holds by the same author on the Architecture Board, each with one or two levels in, because the author got bored with each hold. It is far far better to have one completed hold than five unfinished ones lingering on the Architecture board, or in endless beta. Two holds at once? Unless you're an experienced and published hold builder, just don't do it. It's OK to have one hold that you're adding the finishing touches to (betaing it, voice acting etc) while you build the puzzles for another, but don't try to construct two holds worth of puzzles at once. Why not just combine the puzzles into *one* hold?
Second point - think small. The oft-quoted example of things that go wrong in this case is "Pitiful". It's not a bad hold, but it could have been so much better if it were half the size. A lot of the rooms are repetitions of other rooms, and as a whole it gets kinda boring. I'm speaking from experience here too, my own first hold, Witherwood Castle suffers from this a bit. I was too ambitious, I made it too large, too much, before I was ready. If you set yourself a defined size for your hold, and try to make rooms to fill it, you'll end up making crappy rooms. Don't give yourself a target to work towards, don't say "This hold must have 100 rooms!", it frequently doesn't work. Again, I was guilty of this with Witherwood Castle, I tried to have an entire level themed around each object, and I packed in some crappy rooms as a result.
Consider also that some of the best holds (e.g. the Smitemaster's Selection "The Choice" has a mere 16 rooms!) are short and sweet. Length is not a requirement for quality.
If you've not built a hold before... stick to three or four levels tops. Dischorran's recent hold "Beethro's Brain" is an example of just how good a new Architect can be. It's well presented, the puzzles are original and not repeated, and it's short and sweet.
And that's the lot. That's my method of constructing rooms and holds. It's not perfect, but it works OK for me If you ever want to chat about hold design or construction, I'm usually on the DROD IRC channel (see the forum for details), or when I'm building rooms I'm always on DROD chat. You'll need CaravelNet to access the latter.