The Eye of the Beholder Thread

[anvi]

I don't like RT blobbers but still, EotB and Dungeon Master (i got my hand on a Amstrad pirated* cassette of the latter back when i was in high-school) are classics, millennials spreading their ignorance on videos should be put down, for the greater good.

They are also doing the same thing about comedians, musicians, etc.

 

[Nixheb]

And yet it's mostly forgotten today, while the inferior Eye of the Beholder is not.

Part of the problem is of course that it's not available on Steam or GOG, and not for sale if out of mind for most people, even though free remakes exist.

I wouldn't say that EOB is inferior : it's an excellent variation of the genre, with a different approach. Like wizardry, for example.

 

[Unkillable Cat]

I think the most accurate description regarding a 'Dungeon Master vs Eye of the Beholder'-argument would be that EotB gave people what they wanted - more dungeon crawling in a slightly better package.

DM had good graphics, but very limited animation. EotB had better graphics and slightly better animation (though not by much). DM had an almost alien UI (especially the spellcasting element) while EotB's UI is much more accessible (but still with some rough edges). Finally, having the Dungeons & Dragons-label on it helped make EotB reach a larger audience. Gameplay-wise the games are virtually identical, but EotB has the advantage of coming in when the tech and software is better.

(I'm skipping mentioning Chaos Strikes Back because I still haven't had a chance to play it proper, so I can't tell honestly where it fits into things. But it's there.)

But weren't there any other DM-wannabees released before EotB came along? Yes, but only two really stood out: Bloodwych and Captive. Although both are real-time dungeon crawlers, they're both considerable departures from what Dungeon Master was, and enticed different tastes.

 

[octavius]

The problem is that EoB (and neary all later DM-clones) missed what made DM so great: the physics and being able to interact with the environment, different ways of defeating enemies and opening doors, a detailed hit location system where hurt legs would really slow you down, intricate level design and puzzles.

Later games are superficially DM-clones, but they are really just simplified games with more varied graphics and they added NPC interaction (and MP in Bloodwych's case).
It's like later designers didn't really understand Dungeon Master.

 

[Unkillable Cat]

Oh, I think they did. Possibly more than you think. Even back then you can bet your pants that developers (and publishers) quizzed players about DM and what were 'the good parts' about it, then went back to their offices and designed games that kept the good part and threw out the bad. The simplicity of EotB can almost squarely be put on Westwood's shoulders, who saw how complex both DM was in parts, and then how complex the AD&D-system was, and simplified both to give us EotB.

Captive retained some of what you mention: You can crush monsters both with doors and wall blocks (where applicable), and you can melt some of them in lava (and drown a few of them in the water)... and that's before we get to the anti-gravity physics! Energy can be recharged via batteries, or you can just 'stick' the cursor into an electrical socket and either recharge that way, or use the cursor charge to fire a few bolts of energy at enemies. Captive then goes a step beyond and offers not only damage to body parts, but interchangable body parts. The problem with Captive though, was that it went too far in its environmental interactions, to the point that some of the items are stubborn in their use. The camera is a good example of this.

Bloodwych had the best NPC-interaction mechanic out there... but it served very little purpose overall. You can beat the game without ever using it (outside of the recruiting stage at the start).

 

[Nixheb]

Unkillable Cat On EOB2, there are names 'ALLABAR', 'WILLIAM', 'CIRCE', and 'APHRODITE' in the executable file. If I remember well, I saw those names on some game screenshots. Do you have any idea if those are playable characters ?

 

[Unkillable Cat]

I don't know. I don't recall coming across those names myself.

 

[Black Plague]

Good point Nixheb I went and checked my disk images (v1.6, which was preceded by at least one other version).

The oldest date I could find there was February 18th 1991, and the newest date was April 29th.

So it looks like the first version (v1.4?) was sent out to the masters in February, but reviewers got v1.6 or even v1.7.

I'm holding the original disks in my left hand right now. Do you want me to check the dates?

Spoiler

I wish I had a working disk drive.

 

[Zed Duke of Banville]

I wish I had a working disk drive.

I have working Amiga disk drives but haven't kept a PC floppy drive.

 

[octavius]

D(on't)F0(rget):

 

[Delterius]

I wish I had a working disk drive.

I have working Amiga disk drives but haven't kept a PC floppy drive.

years ago i was reading a conversation right here on the codex

first worlder: 'oh man i can't find a diskette reader anymore'

so i looked sideways and there it was, my computer 3 cpus ago with its gloriously never used diskette reader from third worldian 2000s.

 

[anvi]

Cassettes were amazing, video and music. Floppies can suck my balls though. Abort Retry Fail

 

[affablePeafowl]

This seems to be the most active place for EOB discussion online, so I made an account to ask some things about the game's mechanics and hopefully solicit the wisdom of folks—like Mr. Cat—who have mapped quite a bit of the proverbial terrain, in terms of what the game is doing under the hood. This will probably be a huge wall of text so I'll try to keep the perfunctories brief.

Half the manual is dead wrong, most of the PnP mechanics weren't really implemented, all longtime players know this. I've been collecting a ton of data and running it through a binomial hypothesis test calculator online to answer a few questions I've had, since nobody's ever cared enough to test super granular stuff, and it would mostly be just confirming stuff folks already know, rather than finding surprises. E.g. I can pretty much guarantee the Paladin has no passive Protection aura, like we all already knew, and that saving throws don't really seem to be conditional on race (or at least, there's no difference between a 16 CON Paladin or a 19 CON Dwarf Fighter, if a Spider hits them it seems to be a 50/50 chance of getting poisoned - but much higher for Anya and my Elf mage, which I can't explain quite yet. Oh, and Luck Stones do absolutely nothing for saving throws; again, nothing new or interesting.)

Then I got curious about the supposed dual-wielding penalty for non-Rangers; it's already been seemingly debunked, but I wondered if it might exist in some form. I did a little experiment and got some very odd data. I used the respawnable Wisps in EOB2's Silver Tower. They apparently have the best AC in the trilogy, so I figured they might give me the best data, since theoretically any THAC0 differences would be more pronounced with them. Their AC is -8 according to the Monster Manual, I'm making a bit of an unwarranted assumption that the devs also gave them -8 AC in the game (but the results do seem to support that value). The two characters I used were a Lv9 Paladin (18/78 STR, 17 DEX) and a Lv9 Dwarf Fighter (Keirgar from the first game - 18/92 STR, 12 DEX). I wanted them to have the same THAC0, so they had +3 Short Swords in their main hands (Slicer from EOB1 and the nameless +3 in the Silver Tower) and +2 Short Swords in their off-hands (Sting from the Catacombs, the unnamed +2 from the Silver Tower).

They both get a -2 THAC0 bonus from STR, and at Lv9 they're at 12 base THAC0, so accounting for their weapons, both characters have a main-hand effective THAC0 of 7 and an off-hand effective THAC0 of 8 (assuming the null hypothesis is true and there is no penalty whatsoever for dual-wielding). The Wisps, then, require a main-hand roll of 15 to hit and an off-hand roll of 16. In other words, I should expect both characters to hit 30% of the time from the main hand (6 of 20 possible rolls) and 25% of the time from their off-hand (5 of 20).

I casted Haste and Protection from Evil 10' to speed the process up. No reason to think this would affect the results, but if anyone sees any gaping holes in my methodology, by all means point 'em out. Paladin attacked 124 times (62 with the +3, 62 with the +2) and Keirgar 122 (61, 61). I tracked the results by both hand and character (God bless grid paper). Of the 246 attacks, 66 landed: 26.8%, perfectly in line with the expected 27.5% (average of the 30% and 25% expectations by hand). Damn near the crest of the bell curve of expected results, per the hypothesis test calculator, so I'm confident in my sample size, and from that broad view it certainly doesn't seem like there's any dual-wielding penalty.

But... the results get REALLY quirky if you break it down by main vs. off-hand, and by Paladin vs. Keirgar. I'll wait until my noob-post gets approved before I post the specifics, no reason to make this wall of text twice as long if I don't have to, but suffice it to say I'd love to get some feedback from smarter folk than me. Basically the results are the polar opposite of what you'd statistically expect, and the off-hands with their +2 swords actually outperformed the main-hand +3 ones, by about what you'd expect from a difference of 1 THAC0 (off-hand's hit rate was ~5% higher). Likewise, Keirgar underperformed in both hands by about the same rate, as if his THAC0 were one point worse. So maybe there's something to the "DEX mitigates dual-wielding penalty" thing, but that doesn't explain the off-hand's superiority across both characters.

I know it sounds like variance/sample size issues, but there's a weird, very rigid symmetry to the quirks. I'll get into that when I post the actual numbers. In the meantime, I'm glad to see folks are still into the trilogy! (by which I mean EOB1 and EOB2).

 

[Unkillable Cat]

Welcome affablePeafowl . I'm a little surprised that this seems to be the most active EOB-discussion online.

Anyway, to get into what you're saying:

# At least half the manual is wrong, and even the cluebook is off in places. Having a 'primary stat' higher than 16, for example, is supposed to give a 10% bonus to XP. (It doesn't.)

# I have no data to even suggest that saving throws are in the game. If they are, they don't follow the AD&D rules, except only loosely.

# Good to know about the Protection-aura! I had been wondering about that myself, but never bothered to do the legwork to test for its presence.

# I also haven't done the legwork with the dual-wielding weapon penalty, so I can't help you there, but what you're posting is very interesting.

# What's really been cooking my noodle, is the monster Magic Resistance. I have the hard data from the game files lying around here somewhere... except either I'm reading it wrong or it just doesn't add up. They seem to follow the Magic Resistance-value of these same monsters in the AD&D Monstrous Manuals.

 

[affablePeafowl]

That good ol' 10% XP bonus that one was my introduction to the BS in the manuals, since it was the easiest to disprove. It's mildly frustrating that they at least acknowledged the one error - racial weapon bonuses - in the data card, but then when EOB2 came out they just copy/pasted all the same errata word for word into the EOB2 manual, without correcting it.

I'm gonna double-post and put all the dual-wielding stuff on its own, hopefully nobody'll mind since that post is gonna be long enough as it is. But just for the sake of getting some hard data on record, I'll run through the Paladin's aura numbers and saving throws real quick (which I'm definitely with you on, I don't think they're implemented in the structure we know them as.)

Let a Displacer Beast (15 THAC0) attack my Paladin-led party 120 times and recorded the # of hits. Front-liners both had -4 AC; per the null hypothesis (H0), assuming the Paladin's aura is NOT implemented, the D-Beast will hit on an attack roll of 19 or 20 (15 + 4 AC = 19 to hit), or 10% of the time. The alternative hypothesis (H1) is that the Paladin's Protection aura WAS implemented, meaning the Displacer Beast would only hit on a natural 20, or 5% of the time, regardless of the magnitude of the aura's effect (this is why I set my frontliners up with -4 AC).

The number of hits was 14 of the 120 attempts, obviously well in line with the null hypothesis's expected total of 12 (120 * 0.10). Indeed, per the calculator: The probability of exactly 14 hits in 120 trials is 9.45%. The probability of exactly, or fewer than, 14 hits is 78.18%. The probability of exactly, or more than, 14 hits is 31.27%. H0 holds.

So the last gasp for the Pally aura was the question, is 14 an inordinately high number of hits if H1 WERE true? Inputting the data into the calculator as follows, if...
n=120 (# of trials), and
K=14 (# of successes), and
P=0.05 (expected probability of success, given alternate hypothesis—innate Protection from Evil aura means D-Beast only hits on a natural 20)

The probability of exactly 14 hits in 120 trials is 0.18%. The probability of exactly, or fewer than, 14 hits is 99.90%. The probability of exactly, or more than, 14 hits is 0.28% (.0028). Having a Paladin in the party seemingly has no measurable effect on an Evil attacker's accuracy.

And the "saving throws" (or, more accurately, testing resistance to poison against Giant Spiders:

TEST 1:
Front rank = Paladin (15 WIS, 16 CON, 7 AC) & Keirgar (15 WIS, 19 CON, 7 AC). No Luckstones equipped. Let Spider hit both characters 60 times. Paladin was poisoned 27/60 times; Fighter was poisoned 28/60 times.

TEST 2:
Identical conditions, but with Luckstone Medallions equipped on frontliners. Let Spider hit both characters 20 times. Paladin was poisoned 7/20 times, Keirgar was poisoned 10/20 times.

Conclusion: high-CON dwarves aren't inordinately resistant to poison, and Paladins certainly aren't "immune to all disease," though it's not entirely out of the question that these passive resistances were implemented, and the game simply applies this same resistance to both Paladins and Dwarves (e.g., if class == paladin, elif race == dwarf, resist = true). Probably should've collected more samples for Test 2 but feeling pretty confident that they do fuck-all.

TEST 3:
Revived Anya (11 WIS, 12 CON modified in ASE, 6 AC) and put her in the front row along with my Elf Mage (10 WIS, 15 CON, 6 AC). Let Spider attack both characters 20 times. Anya was poisoned 16/20 times and Mage was poisoned 12/20 times.

Conclusion: Fuck if I know. Probably assed this experiment up, if true probability of poisoning = ~50%, 20 samples is likely too low. Mage's result is in line with Test 1 and 2 but Anya seemed incredibly susceptible to poison. Potentially based on level (Anya starts at Lv4, well behind my party at that point) but sample size likely too low to infer statistical significance.

In hindsight I wish I had been less sloppy with that one. I didn't even record my party's levels, but fortunately I have the save: Paladin + Keirgar were Lv8, Mage was Lv9. I was recording WIS because the manual talks about spell resistances based on WIS, but that almost certainly wouldn't apply to poison and of course we know it probably isn't implemented to any degree. I modified Anya's CON from 16 because I wanted to test different values, but that just ended up letting way too many variables into my very-limited sample sizes in the last test. If I had to guess, there might be a half-assed saving throw table based on level and class (class meaning Warrior, Priest, Rogue and Wizard) but whether there are different STs for different effects (e.g. Mind Flayers, Mantises, various Beholder eyestalks) I have no idea, you'd know better than I would. Instinct says not bloody likely, it's hard to imagine they printed the manual full of simple mechanics that weren't implemented, but they didn't mention a robust Saving Throw system they did implement (ha). Anyway it's info that's out there now, for folks smarter than me to use one day, maybe.

 

[affablePeafowl]

As for the rest of the dual-wielding data: as shorthand, I'll label the four hands as PM, PO, FM, FO (Paladin Main, Fighter Off, etc.) The null hypothesis/expected results again for clarity: 30% from PM and FM, 25% from PO and FO. I should throw out a quick disclaimer, I'm probably not being very scientific by running the tests on specific chunks of the data - I'm basically shrinking the sample size to a fourth of its size by dividing the data like this. The weird symmetry I mentioned is the only reason I'm wondering about these results, so I'll get to the point. The actual data:

PM: 17/62 (27.4%, -2.6 points vs. expected; well within a standard deviation if expected P = .30)
PO: 20/62 (32.25%, +7.25 points vs. expected; still not terribly anomalous on its own, but this result is outside a standard deviation if expected P = .25)
FM: 13/61 (21.3%, -8.7 points vs. expected, again possibly due to variance, but this is WELL beyond a standard deviation, closer to 2 SDs. Four fewer successes than PM in only 1 less trial. You'd expect this result from a THAC0 two points worse than the actual THAC0.)
FO: 16/61 (26.2%, +1.2 points vs. expected. This, along with PM, are the only hands that behaved as they should've, statistically speaking. But again, precisely four fewer successes than PO despite similar sample sizes - I'll get to that later.)

And for reference, how anomalous these results are according to the calculator, which is the meat of this whole experiment after all (no reason to break down PM and FO, since they're pretty much at the top of their bell curves of expected outcomes):
PO: The probability (p) of 20 or FEWER successes in 62 attempts, given expected frequency of .25, is .926 (i.e. 92.6%). The p of 20 or MORE successes in those conditions is .122.
FM: The p of 13 or MORE successes in 61 attempts, given expected frequency of .30, is .951. The p of 13 or FEWER successes in those conditions is only .087.

The best theory is still simple variance, but if you zoom out a little bit and look at it just by hand or by character - only halving the sample size, which should still be reliable - the anomalies don't exactly go away.

Together, the characters hit a collective 30/123 main-handed attempts (24.4%). Given that sample size and an expected frequency of .30, the probability of a total as low, or lower, than 30 is .102. The characters hit a collective 36/123 from their off-hands (29.3%). Given that sample size and an expected frequency of .25, for a total of 30 or higher, p = .161. This is the most noteworthy finding IMO. It's hard for someone as statistically-illiterate as me to know quite how to translate this data into a significance test, since I wasn't testing for this specific data, but AFAIK the null hypothesis would hold perfectly well for those p-values - i.e., they're not a big enough anomaly to reject the null hypothesis (that there's no penalty for dual-wielding).

It's just that weird rigid symmetry that confuses me. As y'all can probably infer, the results by main hand vs. off hand align nearly perfectly with their expected bell curves... that is, if you expected a frequency of 25% for the main hand, and 30% for the off-hand. In which case the p in both directions, for both hands, would be about 50/50 (near the top of the bell curve), instead of the 90/10 outliers. That's the odd rigid symmetry - it could definitely be variance, but if so, the cookie crumbled in exactly the perfect way to confuse the hell out of me. It's like the main hand has a +1 THAC0 penalty... but the off-hand has a -1 THAC0 bonus.

In terms of character vs. character, the data is less extreme, but it does leave the door cracked a bit open for the 16+ DEX bonus to dual-wielding, so I'll run through it. The Paladin hit on 37 of his 124 attempts (29.8%). The fighter, Keirgar, hit on 29 of his 122 attempts (23.8%). The null hypothesis holds for both characters given an expected frequency of .275. Neither of these results are outliers of anything approaching statistical significance. But if you were to expect 12-DEX Keirgar to suffer a penalty of +1 THAC0 in each hand, these are almost precisely the results you'd expect (four fewer successes from each hand, -6.1 points behind Paladin from the main hand, -6.2 points behind him from the off-hand; you'd expect a five-point gap given a difference of 1 THAC0 - there's that weird symmetry again.)

To recap, the overall totals firmly support the null hypothesis, so I'm inclined to think I'm just seeing patterns in noise. We'd expect the total accuracy to be around 27.5%, which it is. We would also expect one hand to succeed ~30% of the time, and the other to succeed around 25% - which did happen. But the +2 off-hands were the one that hit ~30%, and the +3 main hands were the ones that hit ~25%. The fact that using the better, bigger sample sizes only highlights the anomalies, instead of mitigating them, has me really confused. But take all this with a grain of salt, I won't pretend to have any sort of statistical expertise, and there might be MAJOR flaws with how I'm comparing the samples across different levels of granularity.

More experimentation would be ideal, which was definitely the result I feared the most. I should probably repeat the trial, and do a separate one with both characters kitted out sword-and-board, to see what I get from that. But that might be something for a rainy day, since the null hypothesis does seem to hold despite all the oddities. There's no penalty for dual-wielding... probably. Again, if anyone smarter sees where I screwed up the test or where I might be misinterpreting the data, feel free to educate me!

 

[affablePeafowl]

Quick follow-up with dual-wielding test 2.0. Made an EOB1 party with a Human Fighter (12 DEX) and a Human Ranger (18 DEX) in the front and took five minutes to run to the skelly jail in Level 2. Saved the game and wailed away on the skeletons for a while. Both characters were identical aside from class and DEX (both had 13 STR and were at Lv4, for THAC0 scores of 17). Made sure the Fighter's DEX was below the 16 threshold for the manual's supposed bonus to dual-wielding (or rather, mitigation of the supposed penalty). Gave the Ranger max DEX to ensure the effect would be maximized, if present at all—though the Ranger should be exempt from the penalty entirely, there was no reason not to make conditions as favorable for him as possible, since my goal was just to detect whether or not an advantage exists at all.

Both characters had a regular Mace in their main hand and a regular Short Sword in their off-hand. I attacked 72 times with each weapon. The results seem pretty conclusive in that they definitely don't suggest any Ranger bonus or Fighter penalty to dual-wielding:

Total: 106/288 (36.81%)
Fighter: 52/144 (36.11%)
Ranger: 54/144 (37.50%; +1.39 points relative to Fighter, about one-fourth the magnitude you'd expect if there were a difference of even 1 THAC0)
Main Hands: 55/144 (38.19%)
Off Hands: 51/144 (35.42%)

Fighter Main: 31/72 (43.06%)
Ranger Main: 24/72 (33.33%)
Fighter Off: 21/72 (29.17%)
Ranger Off: 30/72 (41.67%)

Didn't bother with hypothesis calculators this time, doesn't seem to be any reason for it based on the data. Probably won't bother seeing what kind of data I get from sword & shield either since, well, even in the unlikely case it has a -1 THAC0 advantage, I'd still rather have two dual-wielders up front.

These results definitely make me inclined to dismiss the degree to which off hands outperformed main hands in the last test. Here, with a bigger sample size and more variables controlled for, the results cluster pretty tightly by character and by hand (the N=144 samples). When you halve that sample size and look at them by each individual hand, the Ranger's off-hand outperforming the Fighter's off-hand by ~2 THAC0 seems significant, until you see that the Fighter's main hand outperformed the Ranger's main hand by roughly the same degree. A bit disappointing: from a pure meta-gaming perspective, taking a Ranger is indefensible. On the bright side, all the Warrior classes are obviously excellent front-liners, and you could cruise through the trilogy comfortably with a frontline of two Rangers. So if a new player were going through the games for the first time and wanted a Ranger because they like 'em, or for RP reasons or whatever, I'd still say go for it.

But yeah, in every conceivable aspect, the Ranger is just a Fighter that levels up more slowly.

Since we already knew that, perhaps what's more interesting is that Skeletons almost certainly have a better AC than the 6 or 7 they should have according to the Monster Manual. The results seem to cluster around 35% to hit, i.e. a roll of 14 or better, which given my 17 THAC0 would be an AC value of 3. Maybe the devs wanted them to be a bit more threatening given their measly HP, but eh, who knows?

 

[octavius]

I enjoyed EoB back in the days, but when I tried to replay it some years ago it just felt wrong, shoehorning AD&D into a DM-like game. Even the first fight against kobolds I felt they hit far more often than they should.

Sometimes ignorance is bliss.

 

[affablePeafowl]

# What's really been cooking my noodle, is the monster Magic Resistance. I have the hard data from the game files lying around here somewhere... except either I'm reading it wrong or it just doesn't add up. They seem to follow the Magic Resistance-value of these same monsters in the AD&D Monstrous Manuals.

AFAIK this is the only other mechanic that hasn't been researched in-depth, that I've seen anyway. The impression I got, from a Let's Play IIRC, is that it's as you said, they just use their PnP values (though Westwood seemingly wasn't 100% faithful in that regard, Skeletons are harder to hit than they should be and, as octavius mentioned, Kobolds definitely seem to hit too often). If I can figure out a way to test and get a rough outline of the general shape of the MR system, the community would have more or less everything figured out, seems like the last white whale is MR and resistances in general, in whatever form they were implemented in the game. The MR percentage is just implemented as a flat probability that a spell will have no effect, right? Skeletal Lords supposedly have a 90% MR and 10% success rate with my spells feels about right (and Mind Flayers... just beat EOB2 for the first time in years and christ I forget what a pain in the ass that Illithid horde before Dran is.)

The data in the game files you have, is it in hex? Or is there some magic utility these days that can translate that stuff into high-level code?

 

[Desiderius]

I wish I had a working disk drive.

I have working Amiga disk drives but haven't kept a PC floppy drive.

Poor old 8-track feeling left out.

 

[Desiderius]

I enjoyed EoB back in the days, but when I tried to replay it some years ago it just felt wrong, shoehorning AD&D into a DM-like game. Even the first fight against kobolds I felt they hit far more often than they should.

Sometimes ignorance is bliss.

My main memory of playing EoB the first time was that I thought the game was about to be over then the wall color changed. I'd just discovered level four. I missed out on DM but game was kind of huge for its time. A lot there.

 

[affablePeafowl]

On an entirely different note, has anyone ever come up with any theories to explain the "Thirteen/Proper Sequence" hallway in the Mantis Warrior level, with the levers and teleporters? The solution to the teleporters is simple, but the levers don't have anything to do with it, and per EOB Explorer they aren't linked to any function at all, i.e. they're purely decorative. The "Thirteen" and the Cluebook's wink-wink-nudge-nudge about it being a problem with a "binary solution" suggest that the player is meant to use the levers to write 13 in binary (which google tells me is 1101). But there's only seven switches and you'd presumably expect eight or four, if you were computer-savvy enough to figure out what the Cluebook was getting at. So I never put much stock into that until I just realized there's a hidden switch in the Special Quest room (where you put Kenku Eggs in the alcove), on the map it's aligned on the same Y-axis as the "Thirteen" switches, only three squares away from that hallway horizontally. The dumb conspiratorial part of my brain wonders if that's supposed to be some kind of 8th switch.

The secret switch doesn't have any apparent effect. It simply sets Flag 01, which itself seems to do nothing—if you step in front of one of the shelves, the game checks if Flag 01 is set and, if not, runs a subroutine at $0370 to remove the walls around the squares in front of the alcoves where the Mantises stand. Except there aren't any walls there to begin with. Flag 01 also shows up if you step into one of those squares:

$09A7 IF TRIGGER_FLAG = 16 (on_throw_item)
$09AB ELSE_GOTO $09B1

$09AE TURN MODE = $F5 (item), AMOUNT = 1

$09B1 IF LEVEL_FLAG (1) = 0
$09B6 ELSE_GOTO $09BC

$09B9 GOSUB $0370

$09BC END

So, if an item is thrown from/into those squares, the game runs the TURN function, which is what re-directs items in midair. And again, checks for Flag 01, and if it isn't set, runs the same subroutine. Except items you throw don't seem to turn at all, so I reckon I'm just misinterpreting the code wildly. TURN's AMOUNT parameter is, AFAIK, just a 0-3 value that determines which direction to re-direct the item. Two of these squares have the AMOUNT set as 1, as above, and the other two have it set as 3. So I guess the hidden switch has something to do with the Mantis room, something that they didn't have a chance to finish, but I can't fathom what they were trying to do. So even granting that the switch has nothing to do with the nearby teleporter puzzle, its actual function is even more opaque than that.

(edit: Correction, if you throw something north while standing in the vertical axis in front of the shelves, the item you throw will indeed turn north 90 degrees, right before it'd hit the northern wall, and flies into the northernmost alcove. But... why?)

There are so many bizarre things in the game if you look around in EOB Explorer (I won't let myself even get started on the weird +9 darts in Level 9, which are there for god knows what reason, and are apparently supposed to get teleported to the end of the level, but they don't). I get that they didn't have time to finish stuff like the "combination lock" and (presumably) "Thirteen" puzzles, but in that case why not take two minutes to remove the "Thirteen" text entirely?

 

[Unkillable Cat]

Because that's how Westwood rolls: They hide away the unused bits, but they don't cut them out. There are MUCH bigger examples of this in EOB2 and Lands of Lore 1.

The +9 Darts are the darts that are shot out of the walls when you step on the pressure plates in the hallway where you find the dead paladin (who's a Fighter in-game, but his gear suggests otherwise). They're probably given that high a value so that the trap does any real damage against the party.

The "Thirteen/Proper Sequence" puzzle is actually half-true - it requires a proper sequence to be solved (though it's only a two-step process). "Thirteen" is not the only mysterious inscription in the game, there's also that "Chwat"-inscription on the 11th floor.

 

[affablePeafowl]

The +9 Darts are the darts that are shot out of the walls when you step on the pressure plates in the hallway where you find the dead paladin (who's a Fighter in-game, but his gear suggests otherwise). They're probably given that high a value so that the trap does any real damage against the party.

That's just the thing, though - this was always my assumption, but I don't think the trap ever deals more than 3 damage. It just fires ordinary 1d3 darts at you. I was looking through EOB Explorer trying to figure out what the Cluebook note at the end of the level is all about, and noticed the +9 darts seemingly don't have anything to do with the trap. The game just uses the LAUNCHER function to spawn regular darts mid-air like all the other traps:

$1169 REMOVE_ITEM TYPE = $0F (dart), POS = <08,09>
$116D REMOVE_ITEM TYPE = $0F (dart), POS = <08,13>
$1171 REMOVE_ITEM TYPE = $04 (dart), POS = <08,09>
$1175 REMOVE_ITEM TYPE = $04 (dart), POS = <08,13> // coords correspond to where the darts are ostensibly loaded, this is I guess where the game deletes them

$1179 TELEPORT TYPE = $F5 (item), SOURCE = <08,09>, DEST = <02,16> // and I guess this is the game teleporting them to the end of the level, hence the Cluebook's blurb
$117F TELEPORT TYPE = $F5 (item), SOURCE = <08,13>, DEST = <02,16> // ...but the game's already deleted them, so I guess that's why they're not there.

$1185 LAUNCHER TYPE = 236 (item), INDEX = 38, POS = <08,09>, DIR = 2, FACING = 0 // and here's the game just using LAUNCHER to spawn the ordinary darts mid-flight?
$118D LAUNCHER TYPE = 236 (item), INDEX = 38, POS = <08,13>, DIR = 0, FACING = 0 // 38 is their index number, the +9s are in the 434-445 slots.

The darts that hit you even have the ordinary dart sprite, while the +9 Darts have the Drow-themed sprite - I used ASE to teleport into the walls to grab them, and found out after the fact about how you can just pluck three of 'em out of the wall. Looking in EOB Explorer now and item 38 is indeed the plain darts, but their *category* index # is 4, vs. 15 for all the other darts. And in EOB2 4 is a null slot, so I guess that's why they don't carry over. Which is good I guess, darts that deal damage on par with Severious is pretty bananas.

So it's like they pre-loaded these weird unique darts to give the impression that they're loaded in the trap, to people 30 years after release looking at the maps. But 99% of players would never even know they're sitting in the walls, you'd have to move and click pretty randomly to stumble onto the ones that you can just pick up. Maybe True Seeing was originally intended for EOB1, and using it would let users see through the walls and surmise there's a trap. And/or they intended to use the weird +9 darts to make the trap actually threatening that late in the game, but accidentally coded it to fire regular harmless darts. Just Westwood weirdness like you said I reckon.

 

[Null Null]

Unkillable Cat On EOB2, there are names 'ALLABAR', 'WILLIAM', 'CIRCE', and 'APHRODITE' in the executable file. If I remember well, I saw those names on some game screenshots. Do you have any idea if those are playable characters ?

Could they be the names of the default party?