For the most part, I stick with looking at low level adventure modules because the vast majority of my actual play is on the low end of the scale. However, I like to look at higher level adventures from time to time, whether to consider how I could adapt them down to my needs or for general inspiration. Of late, I've been looking at I3: Pharaoh.
It's been a slow process because I keep getting so fed up with the actual contents of the merely-38-page PDF that I have to put it aside. At this moment, I've only made it about halfway through (up to the end of the maze), so there's a chance that it changes for the better in the last three levels of the pyramid.
On the other hand, there is some good stuff in that module. For instance, the backstory of Amun-Re's curse is a nice bit of irony (asshole pharaoh exploits his people to build a theft-proof pyramid full of wealth for the afterlife, so a divine curse binds his spirit to the mortal world until his tomb has been plundered), and unlike many cases of interesting backstories in published adventures, there's a reasonable method to actually communicate that information to the players before they enter the dungeon. It'd be a shame to throw away those moments just because the module's general style and implied implementation grates on me. Instead, I'd like to try going through it (insofar as I've been able to read it to date), picking out what I like and thinking about what I'd change about the parts I dislike.
To begin at the beginning, the way that the adventure starts is utterly ridiculous. By decree of some douche bag ruler, the PCs are condemned to wander the ash desert beyond the mountains as punishment for having played a prank on a wizard and then set up a fight between the wizard's "maid" and "wench". One of the NPCs explicitly says that it doesn't matter if the PCs actually had anything to do with it because they're easy targets to exile without upsetting any locals, and another chips in that it's also an excuse for the ruler to claim he's doing something about their recent problems with raiders from the desert. Attempts to cross back over the mountains bordering the desert are always turned away by a large force of guards, and the whole "raiders from the desert" thing doesn't go anywhere in the actual desert (at least as far as I3 is concerned), so I can only conclude that the "raiders" are actually a False Flag Operation that the ruler uses to give himself leverage for whatever he wants.
On the plus side, after this whole farce is wrapped up and details are given for the mounts and supplies the PCs are provided with, there's a table of rumors about treasures and dangers in the desert that the PCs picked up from hearing chatter as they were being escorted to its border. The rumors alone can work as a starting point for the adventure, so I'd definitely pull them out to be more accessible to the PCs. If the players want to run off into the desert after hearing about a golden palace or a theft-proof pyramid of treasures, I see no reason to stop them. I'd get rid of rumor 4 and maybe rumor 5, as both relate to sequel adventures that I'm not too interested in at this point, but a rumor about the desert falling away into a bottomless sky is crazy enough to sound intriguing. For rumor 10, I'd want to come up with something about a wish-granting genie trapped in a lost temple. With that cleaned up, there's no need to force the adventure's standard opening. I'd probably still keep it in mind as something that may happen if the PCs draw too much negative attention from the aristocracy, but I'd change the false accusations to something more believable. In that way, it could plant seeds of rebellion against the ruler in the players' minds. That's not supported by I3, but I've got no problem with the PCs using the treasure they gain in the desert to bribe the guards/"raiders" into being on their side as they work to either blackmail or overthrow the local government.
Unfortunately, the desert itself is mostly trash. It's a hex map of 228 or so two-mile hexes, of which eight hexes have actual content (and that's being generous enough to consider an old campsite with nothing special about it and hex that just says "the trail ends here" as content). The terrain is also monotonous: all flat desert aside from six hexes of hills (lacking any other content on them) and fifty-six hexes of "sinkholes" (aka dry quicksand, and again lacking any other content, so they're just a "roll dice to not get fucked" tax). I like having some empty space in hex maps/dungeons (actually, I prefer to make empty space in hex maps by randomizing whether the hex's keyed encounter happens, but I digress), but a 96.5% rate of empty space is absurd.
The random encounters while wandering the desert do little to improve matters, either; of the sixteen possibilities (10-in-12 "normal" and a 2-in-12 chance for 1d6 "special"), six (37.5%) are unintelligent critters, two (12.5%) are weather events (which I like having on wilderness encounter tables, so I was happy to see them included), and three (18.8%) are other environmental effects (seeing a mirage or a distant plume of smoke; nice touches, again), thus only five (31.3%) give any chance of meaningful interaction. Oh, actually, even that's being too generous, because the pegasus-riding airlancers "will land to investigate the party but will not trust them and will not disclose their place of origin under any circumstances[...]will quickly depart[...]will offer no aid to the party and certainly will never take any party member with them when they leave". Instead of giving an encounter that's all about what it isn't, would it have been so hard to give an encounter that could be interesting?
To start with, I'd ditch the worthless airlancer encounter and compress the other entries that come up multiple times (sandmen, dust diggers, and thunderherders) into single entries of variable size. That leaves us with six "normal" encounters and five "special" encounters. Normally, I'd rather fit the whole encounter table into a single roll instead of a stepped roll, and I'd also include the chance of having the hex's keyed encounter as part of that (see Justin Alexander's hex encounter approach for details), but all that specific mechanical implementation is beyond the scope of what I want to talk about here. I3 tries to have a Arabian/Egyptian feel to it, so those five vacant encounter spaces should be filled with appropriate entries for a mid-Eastern flavor. Jackals, hyenas, cheetahs, lions, tigers, boars, rhinos, elephants, and eagles were/are all native to that environment, so any of those could work nicely to imply an ecology with the dust diggers, thunderherders (which I'd replace with fallow deer), and giant tarantulas. Since two of those are carnivores, I'll go with boars to add another herbivore. For more mythic creatures, manticores are a personal favorite for a touch of Persian mythology, and I can't resist also adding a roc to the list (which I'd use in place of the purple worm encounter, because not every fantasy desert needs to rip off Dune). That leaves two more "normal" encounter and one more "special" encounter. How about ghouls (reskinned as deformed cannibals/necrophages, to be closer to their original myths) and faeries (reskinned as peri, again closer to their original myths) as normal encounters, and either shedu, djinni, or night hags (reskinned as divs to once again be closer to their original myths) as a special encounter? That leaves me with:
1 - Fallow deer (critter)
2 - Giant tarantulas (critter)
3 - Dust diggers (critter)
4 - Boars (critter)
5 - Sandstorm (environment event)
6 - Acid rain (environment event)
7 - Peri (intelligent beings, benign)
8 - Sandmen (intelligent beings, malign)
9 - Manticores (intelligent beings, malign)
10 - Ghouls (intelligent beings, malign)
11/12 - Special encounter
1 - Roc (critter)
2 - Plume of smoke (environment event)
3 - Plume of smoke (environment event)
4 - Mirage (environment event)
5 - Dervishes (intelligent beings, benign)
6 - Divs (intelligent beings, malign)
Compared to the original encounter table, I've increased the variety while still sticking to thematic material, but the rates of critters/intelligent beings/environmental events are about the same, so it'd probably be worthwhile to drop one critter in favor of more intelligent beings. Looking ahead a bit, the encounter table for the dervish temple includes dwarves, which don't really show up anywhere else in the adventure, so a chance of encountering some Bedouin-inspired dwarves in place of, say, giant tarantulas would be reasonable.
It should go without saying that the hex map also needs far, far more keyed locations. Landmarks, ruins, encampments or lairs, and so forth should be sprinkled in liberally. Keying something into every hex would be ideal, but given that there are over two hundred of them, I'd settle for at least keying half of the hexes as a baseline. Some more variety to the terrain wouldn't hurt, either; rugged badlands, ash flats, cactus forests, or dried riverbeds all come to mind off the top of my head as options to break up the interminable expanse of "desert" (as someone who's spent considerable time in actual deserts, I know they aren't really featureless swathes of endless sand).
To go back to something good in the adventure, I like that the entries for the ruined gates, the fallen obelisk, and the buried statue mention what information can be read from their inscriptions. While I'm a little put off by having only an 30% chance to do so under normal circumstances, I can understand that as an attempt to represent erosion of the inscriptions, and the one inscription that's actually useful for more than background fluff (the tablet within the sunken temple) is translated without any percentage check.
The sunken temple itself is also a tolerable mini-dungeon. Using Arabic words as the passwords for the magical trap in the main room is a little irksome, but it's hardly the most offensive thing that TSR ever published; hell, I'd say it's less offensive than the misogyny at the start to this very adventure. The thing that actually turns me off about the sunken temple is that the main treasure can't be looted without releasing the creature trapped therein (which is fine in and of itself), yet there's not even an attempt at giving an explanation as to why (ah, there's that secure railroading we needed after having so much freedom to masturbate in the empty desert). Still, it's easy enough to fix that by ignoring the part about the treasure being immovable without releasing the creature. Coming up with a reason why it can't is also fair game, I suppose, but then you lose the opportunity to have a mad genie wreak havoc as part of the PCs' insurrection, and we'd all be poorer for that loss.
The oasis that grants a vision before disappearing is cool, as is getting to encounter the wandering spirit of Amun-Re to learn about his curse. He rambles on far more than he should, but it wouldn't take much effort to break that up and impart the information as part of a conversation instead. I would've rather put the spirit on the random encounter table instead of binding it to a handful of hexes, but at least it's there in a way that's not just fiction for the GM's benefit, and it saves a few hexes of other content that I'd need to come up with. The fact that the players can encounter Amun-Re's spirit multiple times and have him repeat himself without recognizing them on each such encounter is a nice touch; it's unlikely, perhaps, but it would help emphasize that he's suffering from a curse if it were to happen.
All in all, the stuff on the desert hex map is decent, and it's mostly the dearth of it in comparison to the size of the hex map that's bad. I still dislike the railroading in the sunken temple, but given that the same trick is used as one of the potential hooks for Death Frost Doom, maybe I'm just a bit extreme in my aversion to that tactic.
The dervish temple is largely a waste of space (as well as being more mildly offensive orientalism). There's interaction to be had there, sure, and the veiled clues about the rest of the pyramid are always welcome. The problem is that the pyramid is only guarded by a twenty-foot-high wall, so it's pretty trivial to bypass the dervish temple entirely even before considering this adventure is allegedly for level 5-7 PCs. The random encounter table for it also raises some questions. The dwarves can be reasoned as other inhabitants of the desert trading with the dervishes, despite their absence from the adventure's desert random encounter table, but the bandits are absent from anywhere else aside from the pyramid's maze so far as I've seen, and the pack of thirty-five giant rats (always exactly thirty-five!) just feels completely out of place. I'd dump both in favor of more dervishes/dwarves/ghouls/peri (the latter two depending on how I'd imagine their relations with the dervishes), break up the dervish entries based on what they're doing instead of encounter size, get rid of the wall around the rest of the pyramid (which does nothing since the pyramid is only normally accessible through one entryway that's behind the temple), and ramp up the militant protectiveness of the dervishes.
As an aside, I don't think the Hickmans understood how writing with complex characters/glyphs instead of an alphabet works when this adventure was written. The dervish temple is the first of a few places where a single, explicitly non-magical "rune" translates into a small paragraph. I get what they were going for, but the execution was lacking. But I digress.
The two dervishes guarding the entrance to the pyramid proper are a cute touch, I guess. They're also weak enough to be nullified by first-level spells like sleep or charm person (regardless of the specifics of the PC party, those spells exist in the game world as something any wizards could have, so defenses should take them into consideration). It's easy enough to remedy by increasing the guard and spreading them out more along the path between the temple and the pyramid.
The trick of the plundered tomb is actually pretty clever. It's got all sorts of secret doors to empty areas, but that's all a ruse to distract from the actual means of progression (unrelated, I wonder if the latter was inspired by C.A. Smith's The City of the Singing Flame). However, exploring it is still useful, since it could be used as a destination to teleport out of the deeper levels without risking appearing in the middle of a crowd. I like it.
What I dislike vehemently, though, and what prompted me to write this whole thing is the start of the maze. The repeated bit of advice to describe the surroundings in terms of relative directions instead of compass directions is worthless since that's clearly how area descriptions should always be framed if you're playing in a way where players have to make their own maps (yes, I'm saying Gygax's descriptions in his examples of play were bad). Far worse than that redundancy, though, are the contrivances to make traveling through the maze challenging. Several parts of the maze are full of perpetual mist which obscures sight completely (fine), fouls infravision (fine), makes it impossible to judge distances (maybe fine, depending on interpretation), disorients the PC (hackles rising), and makes it impossible to count (annoying, but easy to overcome, e.g. having a PC call out "stride!" or some such on each step as they walk through the mist while someone outside of the mist keeps count). Oh, but my favorite part is that all of the corners in misty areas are rounded, so that "characters who are feeling their way along the walls will not notice [turning]". That's not how turning works. It's obvious bullshit just to make a maze difficult.
Here's an idea: don't just have a static maze that relies on contrived nonsense to be challenging! Have corridors where the physical distance in real space doesn't match with the distance traveled to traverse it. Have intersections where certain exits open or close based on other choices. Have thresholds that act as teleporters (so that exiting one room to the left makes you enter another room from the top, for instance). Wizardry: Proving Grounds of the Mad Overlord did the latter two (and more) in 1980, two years before I3 was published. B1: In Search of the Unknown had teleporting/reorienting rooms in 1979, three years before I3 was published. Pretty much all of the traps listed in OD&D Vol 3 were aimed at making players get lost, and that was released in 1974, eight years before I3 was published. Clever means of tricking players into losing their way can lead to rewarding players for noticing that something is wrong with their mapping and investigating/experimenting to figure it out. Now, yes, I can imagine that some use of rope, poles, chalk, and so forth can overcome the bullshit of the mists. My greater problem with them is that overcoming the bullshit does little to enable the players to then use that knowledge to their advantage, whereas tricks like shifting passages/walls, teleporters, corridors that are shorter/longer than they appear, and so on can be leveraged into creative approaches for handling encounters, transporting treasure, and other elements of dungeoncrawling.
In other words, clever mazes enhance other parts of play. I3's maze is just an end unto itself, since it takes a long way around to ending up at "you can't tell what's happening in the mist" with little payoff. Once solutions for its obfuscation are found, they don't really provide opportunities to impact other parts of play; they just let the players get on with the rest of the game.
I'd keep the perpetual mists as a means of obstructing vision, but I'd make notes of a handful of intersections with sliding walls to seal off two of the exits. Since the maze's entrance includes a lever-based trap already, I'd expand on that by having levers in the rooms that otherwise contain morphic monsters which can change the exits that the sliding walls are sealing. Add in a few thresholds at the edges of the mists that act as teleporters (maybe three such pairs in total) and at least one section along each cardinal direction that takes twice as long to walk as it actually is, and the maze ends up being far more interesting, in my opinion. These are hardly the only ways of spicing up the maze, of course, but they're just examples for illustration.
Most of the keyed encounters within the maze are reasonable enough, aside from the morphic monsters (I'm just not a fan of AD&D trappers/lurkers, and mimics tend to be a turn-off unless there's some telegraphing) and the minotaurs (too cliche and against the theme of the other creatures). I can't decide whether the "X" room is funny or silly, but I'd be inclined to keep it as a change of pace to help relieve built-up tension. The note about the doppelgangers trying to split up the PCs using the mists is good, as is the sphinx room's variation on the typical "Knights and Knaves" logic puzzle. The random encounter list is kind of a mess. The dervishes, thieves, doppelgangers, and ghouls are fine as random encounters, but I'd switch out the giant spiders for scarab beetles (to fit better with the pyramid theme) and add fungi and oozes along with more undead in place of the wizard and minotaurs. The list of doppelganger forms is a nice touch, though I'd want them to mimic people that actually show up (e.g. dervishes and dwarves) in order to be more confusing. The boring coin treasures should be replaced (at least in part) with fanciful arts and crafts.
That covers the broad strokes of the changes I'd make for as much of the adventure as I've read so far. There's a fair amount of good stuff in it, and aside from the beginning and the genie in the sunken temple, the restrictions tend to feel more like sensible chokers than railroading, at least on paper. Filling in the desert with enough interesting stuff to make it worth exploring would be a significant endeavor, so I wouldn't do it just for this adventure; I'd save that for running a campaign with wilderness exploration that features a desert (which may or may not include this adventure, but that's another story). Fixing the maze would be where I'd focus most of my effort, assuming the final three levels of the dungeon aren't in need of much more attention than cleaning up random encounter lists.
That all said, I don't foresee needing a pyramid dungeon in my immediate future (I do foresee a need for a lost desert tomb, but I imagine it'll be some time before it actually comes up in play), so I won't get to try any of this out at the table for a while. If anyone happens either to have experience playing/running I3 against which to compare my thoughts or to be thinking about running it soon for themselves, feel free to share your perspectives.