(This continues from my previous post about this French game for the Tandy PC-1 Pocket Computer.)
From Byte, February 1982.
Last time, my two issues with Des Cavernes were
(a.) I only had a “simulator” and not an “emulator” and it doesn’t handle memory like the original PC-1 (or the equivalent Sharp 1211)
and
(b.) Jim Gerrie had a port but it seemed to be buggy.
Unfortunately, as of this writing nobody has dumped the BIOS for a PC-1 or 1211, so (a.) remains an issue, but Jim Gerrie has fixed his code, and the issue it had is fascinating and reflects on the raw ingenuity of early 80s programmers.
Before getting into that (and how the fixed game plays) I wanted to mention a second article from Trace Issue #2, the same one that Des Cavernes was printed in. There’s a page-long introduction to Crowther/Woods Adventure that gives a good window into the French knowledge of adventures circa April 1982.
A comic that goes with the article. The computer is saying “you are alone in a dark room… what should you do??”, with the responses “Oh… lighting…” and “don’t forget it”.
The article announces that the “famous” game Adventure, originally in FORTRAN for a PDP-10 can now run on a TRS-80. (This is referring to Microsoft Adventure, based on the version Gordon Letwin originally wrote for Heathkit.) It is “enough to entertain you for long hours, but in English”. I am not clear if the author was aware of Bilingual Adventure; probably not given it was written for the CP/M operating system.
The article goes on to say…
La disquette “adventure” inaugure un nouveau type de jeu où le participant tient un rôle et vit des aventures dont le déroulement est fonction de ses actes. Aux Etats-Unis, on achète une “disquette” d’aventures comme nous achèterions un “livre” d’aventures. Les auteurs ne sont plus Stevenson ou Conan Doyle mais Scott Adams et Microsoft … C’est le progrès !
…translating as:
“Adventure” inaugurates a new type of game in which the participant takes on a role and experiences adventures that unfold according to their actions. In the US, we buy an adventure “disk” like we would buy an adventure “book”. The authors are no longer Stevenson or Conan Doyle but Scott Adams and Microsoft … it’s progress!
Note the “new type of game” comment. This echoes the same line used introducing Omotesando Adventure to Japanese audiences. Additionally this gives the concurrent idea of “bookware”, that text adventures are about to supplant books, something that would briefly catch the imagination of the real publishing industry before disappearing shortly after.
Then, the publication’s “14 year old tester” named Stéphane (who “perfected his English with this game”) gives some “tips and tricks” for tackling Adventure. One section is on “Phrases magiques, passages secrets, objects énigmatiques” and mentions the magic words XYZZY and LWPI. (LWPI is specifically from Microsoft Adventure, and transports the player to a Software Den that’s only in that version and spinoffs. I really should write about it as a standalone article someday, although the differences are really quite minor.)
So while Folibus had technically just come out in a different publication, it was for an entirely different system and it isn’t like awareness of what an adventure even was would spread out immediately, and Adventure in particular was being played in English, not French, despite the technical existence of one translation.
This means that while Crowther/Woods Adventure was “famous”, Des Cavernes would be for some readers the first real encounter with anything like an adventure game.
Now, I’m hedging with “like” an adventure game because even with fixes in, Les Cavernes is a bit unusual, as it is an adventure-roguelike with everything randomly generated. Other than some language fixes, the big issue Jim Gerrie ran into with his port was with the quasi-random number generator itself.
In a technical sense, there is no such things as a random number generated by a regular computer chip. (It’s possible to hook up radio receivers to use atmospheric noise like the website random.org does, or use some related manner of gizmo; I’m meaning normal traditional computer chips.) The best they can do is apply a mathematical algorithm which provides a sequence which gives the appearance of randomness, and this can sometimes go awry. Usually this randomness gets kicked off by a “seed” of some sort, perhaps taken from the system clock (concatenate hours, minutes, and seconds, for instance, into one number) but it can be given explicitly, as Des Cavernes does.
10 “A”:INPUT “NO.=”;D,”L=”;F : F=4*F
In order to then turn this into something appearing “random”, the original game uses the SIN (sine) function. Quoting from Jim Gerrie:
In the end I realized the problem was also a result of a slightly lower level of mathematical accuracy between the MC-10 and Pocket PC. The game relies heavily on the mathematical accuracy of the Pocket PC and its BASIC. Simply put, it needs 10 decimal digits to come out of SIN, whereas the MC-10 could only give 9. That is because the decimal is multiplied by 100 to give two hole number digits, which are used for combat calculations, plus 8 decimal numbers, which are used to store maze node information for 4 directions of moves. The Pocket PC apparently could give you a number like
90.12345678
Whereas the MC-10 can only give you
90.1234567
The 8 decimal number give 4 groups of 2 numbers which store the node information for the four directions of movement. If the first digit is 4 – 9 then you can go in that direction. If the second digit is odd and there is a monster present, then you will be blocked. If you defeat the monster then you will be allowed to move to a new room by adding one to that digit to make it an even number and then a new SIN number will be generated based on the current number. Since I was missing an 8th decimal digit I made it so that it is simply replaced by a zero.
In other words, using 90.12345678 as an example, it was supposed to give the “random” two digit numbers of
12, 34, 56, 78
but instead was giving
12, 34, 56, 70
due to the pocket computer having one more digit of precision than the desktop computer, and since individual digits matter in terms of generating the rooms, always delivering a 0 was causing an issue.
This put together with other fixes create something resembling a game, although I did find I could still run into a “trap seed”, with a no-win scenario. For example, seed 1111, any difficulty:
Here’s the complete map of the level:
There are 10 rings scattered around (some may start in your inventory). If you have the appropriate ring, you can kill a particular monster. In the seed above, I only had RING #1, which apparently is no good with dragons. (KILL DRAGON just has a response of NO!, or if you prefer the French, NON!)
After mucking about with multiple failures, I found a better seed: 2321234. This starts you with rings 1 and 10, so there are technically 8 more rings to find. I played at difficulty 1 (the choices are 0 through 5); higher difficulties add more monsters so are more opportunities to get stopped.
This comes out to be a mess, but at least it is manageable with some persistence. The important thing is not to treat this like a game of Solitaire where you get one shot at a seed. There are far too many dead-end spots like the dragon room I showed earlier, or long dead-end loops (which I’ll show off in a second). No, what makes this playable (once you find a seed that works) is to keep returning to that same seed and add to the map you made last time.
For the map above (not complete) I “died” around 10 times hitting various impossible spots. The thing to keep in mind is that monsters generally don’t have to be killed (although killing one will get a point), and will only block some of the exits (or maybe even none of them).
Monsters will respawn when you re-enter a room. It may seem at first there is no such thing as re-entering a room — nearly everything is a one-way exit, and the two-ways exits I found I think were by luck only — but the map does create “loops” that will eventually return to sections. This is both good and bad; good in that there is some sort of continuity that makes it feel not like I’m just plotting the output of a spreadsheet, but bad in that the impossible loops I alluded to can happen.
Here I have marked out in yellow one such loop. If you enter into the loop without any way of killing the monsters within (vampire and dragon) you are stuck in a softlock with no way out and need to reset. This is very discouraging if you try a new seed each time, but it feels a little like a “discovery” when repeating multiple attempts on the same seed.
It turns out that this seed does have Ring #6 in a reachable spot, and that ring kills vampires (helpfully — and almost certainly by coincidence — there was a vampire in the same room as the ring so I could test it).
Since monsters respawn and each kill gets you a point, you can technically get an infinite score if you can find a loop that brings you by a monster you are able to kill. I found a perfect such spot, as either ring 1 or 10 (I have no way of knowing which) works on wizards.
Yes, that’s a two-way exit. This seems to be complete coincidence, although I do think the map-making algorithm (which I still don’t fully understand) does have “regions” where it is more likely to loop to a room that’s close than a room that’s far. I might be wrong about that guess, though. Since there are two killable monsters in adjacent rooms, it is possible to get any score at all desired by just hopping back and forth and using the KILL command over and over.
I confess I did not persist to try to get absolutely every ring, but in the end this is only adventure-adjacent; this is more of a strategy game like Wumpus which generates a layout you have to reckon with, and what you reckon with just might be impossible, but there’s nothing more complex than mapping going on (and keeping track of if you’re repeating a room you’ve already visited).
Still, I’m glad I got to play this, as it makes for another terrific example in the roguelike-adventure stash (they all take radically different approaches to how they generate their maps; probably the game most comparable to this one is The 6 Keys of Tangrin) and I got to boggle a bit over the sheer technical achievement: remember, this was on a pocket computer. I could see myself playing this more if the PC-1 was the only computer I had, and it was the only way to get at those new famous “adventures” and experience the “objects énigmatiques” within.
Also, despite the room names being generative off a list of descriptors, some of them briefly felt like real locations. Here, I’ve entered a teenager’s bedroom.
Renga in Blue 