The story of Shadowkeep, even more so than Amazon the odd duck in the Telarium lineup, begins with Sigma Distributing, one of the first big microcomputer hardware distributors in the Seattle area. In 1981 Christopher Anson, a Sigma vice president, sought and received permission to start a new subsidiary to develop original games to serve the growing demand for software for the computers Sigma was selling. Anson’s first two acts were to name the company Ultrasoft and hire a programmer named Alan Clark away from Boeing. Clark became the technical architect of the set of tools and approaches that would define Ultrasoft during their brief existence.
Anson had decided that the best place for Ultrasoft to make a splash was in the field of illustrated adventure games, a nexus of excitement in the wake of Mystery House and The Wizard and the Princess. Like Scott Adams, Ken Williams, and Marc Blank before him, Clark realized that it would be more efficient in the long run to write an adventure-game engine and language that could give designers a bit of distance from the technical details of implementation as well as let Ultrasoft deploy their games to multiple platforms relatively painlessly. Deciding that a little corporate branding is always in order, they named their adventure programming language simply Ultra; the interpreter for each targeted platform UltraCode; their graphics system UltraVision. From the standpoint of the end user, Ultrasoft’s most obvious innovation — or, if you like, gimmick — involved this last. UltraVision could display not just static pictures but also brief animations, which could be used to, say, show the player’s avatar actually walking from room to room. Less obvious but no less significant, however, was the parser, one of the first developed outside of Infocom that allowed more than two words — although, it should be said, it was nowhere near as impressive a creation overall (more on that shortly).
Ultrasoft developed two adventures using the system, The Mask of the Sun (1982) and Serpent’s Star (1983). Both are interesting in their way, more carefully crafted, atmospheric, and thoughtful than was the norm of the time. Serpent’s Star in particular does a surprisingly good job of matching its puzzles to its theme of Buddhist philosophy. But both — and particularly Mask of the Sun — are also riddled with the sorts of unfair elements that were all too typical of their era. And both are fairly excruciating to play under any conditions. Whatever its other merits, you see, the Ultra system is slow. A quick look at the games’ technical underpinnings gives a clue as to why: the Ultra program logic doesn’t appear to be compiled at all, merely interpreted in place. The only blessing of that approach was that it enabled some frustrated adventurers to find the solutions to the more incomprehensible puzzles by code diving.
Ultrasoft first tried to market and distribute Mask of the Sun and Serpent’s Star on their own, but found it tough sledding for a tiny company with mainly regional connections in the professionalizing software industry of 1983. They soon accepted the role of developer only, licensing both games to Brøderbund for publication. After spending most of 1983 working on an ambitious new game, an adventure-game/CRPG hybrid called Shadowkeep written in a new version of their system which they dubbed Ultra II, they found a publisher for it in Spinnaker, who took the largely completed game as a future member of their planned Telarium line.
Looking to find some way to make the game fit with the bookware theme of the line as a whole, Spinnaker came up with the idea of reversing the usual process, of hiring a name writer to adapt the game into a book rather than the opposite. Luckily, they had substantial time to get the novelization done; they signed the contract with Ultrasoft in late 1983, a year before they planned to launch the Telarium line. Spinnaker approached Warner Books, who hooked them up with the reigning king of media tie-in novels, Alan Dean Foster. After making his reputation within the industry ghost-writing the Star Wars novelization for George Lucas, Foster had gone on to do The Black Hole, Clash of the Titans, The Last Starfighter, and Alien among many others. Virtually any big science-fiction or fantasy movie released seemed to arrive with an accompanying Alan Dean Foster novelization. (That’s still true today; his most recent novelization as of this writing is of Star Trek into Darkness.)
Spinnaker furnished Foster with design documents and a copy of the Shadowkeep source code and let him have at it. Those were all he had to go on; he doesn’t recall ever meeting or speaking with anyone from the design team, nor ever actually playing the game. He does, however, recall it as a very challenging project indeed. Being a party-based CRPG in the mold of Wizardry, Shadowkeep has no actual protagonist to speak of — no characters at all, really, outside of the fellow who sells you equipment and the evil demon Dal’Brad who shows up for the final showdown on the last of the dungeon levels. He was thus forced to invent the vast majority of the novel himself, whilst struggling to strike a balance between writing some recognizable analogue to the experience of the game and giving away all of its challenges. He did his usual workmanlike job, handing in a readable little genre exercise for simultaneous release with the game. Tellingly, it’s not until halfway through the book that the heroes enter the Shadowkeep — i.e., reach the beginning of the game.
Said game is… well, it’s really strange. Imagine Wizardry played with a text parser, and you’ve pretty well summed up Shadowkeep. You make a party of up to nine(!) characters. Most of the usual is here: attribute scores, classes and races to choose from, ever better equipment and spells to collect. Oddly missing, however, are character levels and any concept of experience; getting more powerful in this game is strictly a matter of finding or buying better stuff. The dungeon levels are the usual 16 X 16 grids full of traps, monsters, and assorted cartographic challenges. There are some original ideas here. For instance, the positions of the monsters that attack you and those of the members of your party are taken account of to a degree not found in Wizardry, adding some strategic depth to the experience. You likewise have more combat options than in Wizardry; in each round you can choose to forget defense and attack twice, or to just parry, or to attack once while not totally neglecting defense. And certainly the full-color graphics, which feature occasional examples of Ultrasoft’s trademark animations, are much better than Wizardry‘s wire frames.
Still, Shadowkeep mostly just makes you appreciate all the more how well Wizardry does the dungeon crawl. The game replaces Wizardry‘s hot-key interface with, yes, a text-adventure parser. You literally just type what you want to do: “OPEN DOOR,” “GET THE TORCH,” “CAST THE LUMINANCE SPELL,” “LIGHT THE TORCH AND PREPARE THE SWORD,” “PUT THE WAND OF TRAVEL IN THE CHEST.” Sounds fine, right? Well, what sounds fine in the abstract doesn’t work so well in practice. You must now type “F <RETURN>” (for “FORWARD”) instead of just “F” every time you want to walk forward a square in the dungeon. This may seem a minor thing, but consider that you’ll be entering this command thousands and thousands of times in the course of playing the game. That extra keystroke thus means thousands and thousands of extra keystrokes. And that’s the tip of the iceberg; this game is death by a hundred such small cuts. Commands by default are carried out by the leader of your party, who is not even a character you select but merely the one with the highest Leadership attribute score. Having someone else do something requires that you prepend her name to the command (“NAOMI GET THE TORCH AND GIVE IT TO REB”) — yet more tedious typing.
And the parser, that focal point of the whole interface, is at least as exasperating as the mainline Telarium parser. Like Byron Preiss Video Productions and many others at this time, Ultrasoft chose to take a profoundly misguided approach to this most critical piece of their engine. As described in an article in Softline magazine:
Ultrasoft’s parser is based on concepts in artificial intelligence. In any given message, it eliminates words that don’t make sense and attempts to make sense out of words that are relevant to the situation. This method frees the player from the verb-noun format of the typical adventure’s input.
In other words, the parser pretends to be smarter than it is by simply throwing out anything it doesn’t understand and doing what it can with the rest. This approach may “free the player from the verb-noun format,” but it also guarantees that complex (and often not so complex) inputs will be not just rejected — which combined with a proper error message is at least a form of useful feedback — but misunderstood. Far from making the parser seem smarter, this just makes it seem that much dumber and that much more infuriating. It leads to situations like that in the Byron Preiss games where any input containing the word “LOOK” anywhere within it causes the parser to dump everything else and print a room description. In Shadowkeep, typing “NAOMI CAST CURE SPELL ON REB” leads her to cast it away into the ether — that “ON REB” was a bridge too far, and thus ignored. Such a system fails to recognize that at least 95% of the time those extra words are not just stuff the player tacked on for the hell of it (who wants to type more than necessary under any circumstances?) but essential information about what she really wants to do.
To play Shadowkeep is to constantly wrestle with the interface. After playing several hours there are basic tasks I still haven’t figured out how to do — like how to cast a cure spell on someone outside of combat, or how to just get a list of the spells a certain character knows. And, like Ultrasoft’s earlier games, Shadowkeep is slow. Every step in the dungeon seems to take an eternity, and as for more complex action… forget about it. Playing is like wading through molasses with shackled feet.
The rewards for all the parsing pain are relatively slight: a handful of logic- or object-oriented puzzles on each level that can perhaps be a bit more complex than they could be under the Wizardry engine. Needless to say, they aren’t worth the rest of the trouble, making Shadowkeep something of a lowlight in the long, chequered history of adventure/CRPG hybrids. Which is a shame, because Shadowkeep‘s dungeon levels do show evidence of some careful craftsmanship and, as noted above, there are some good, original ideas on display here. Shadowkeep is a perfect example of a potentially worthy game destroyed by horrid interface choices. And I mean that literally: if the game isn’t outright unplayable (some patient souls have apparently played and even enjoyed it), it’s closer than I ever need to come to that adjective.
Ultrasoft was already in the process of fading quietly away by the time of Shadowkeep‘s late 1984 release. They never managed to port the Ultra II engine beyond the Apple II, leaving Shadowkeep without that all-critical Commodore 64 version. Spinnaker toyed with doing the port themselves, even announcing it as coming soon on various occasions, but I see no reason to believe that ever happened. (A Commodore 64 version has been a semi-mythical White Whale in collecting circles for many years now, but, despite some anecdotal claims and remembrances, no one has ever produced an actual working version to my knowledge.) The lack of a Commodore 64 version and the underwhelming nature of the game itself combined to make Shadowkeep the least successful — and, today, rarest — of all the Telarium games. Alan Dean Foster’s book, while no bestseller itself, appears to have sold far more copies on the author’s name recognition and its $3 (as opposed to $35) price tag.
Shadowkeep consists, like most of the Telarium games, of four disk sides. In this case, however, all four sides are written to during play to preserve the current state of the dungeon levels; the player is expected to copy her originals before beginning. Most of the copies floating around the Internet contain the residue of the previous players in their dungeons. Thankfully, however, reader Peter Ferrie has provided me (and thus you) with a completely pristine set just waiting for you and only you to leave your marks upon them. If whilst playing Wizardry or Bard’s Tale you thought to yourself that this game would be even better if it played a lot slower and had a parser, you’ve just found your dream CRPG. All others should consider this one a subject for historical research only.
And on that less than stellar note we’ll be moving on from Telarium for a while. My final reckoning of their first five releases is: two worthy efforts (Dragonworld and Amazon); one could-have-been-a-contender (Fahrenheit 451); and two total misfires (Rendezvous with Rama and Shadowkeep). Not a horrible track record on the whole. We’ll see if they learned any lessons in time for their last few games down the road a ways. But next it’s time to get back to the big boys in the field, and tell the rest of the story of Infocom’s very eventful 1984.
(In addition to the sources listed in my first article on bookware and Telarium, I also referenced for this article a feature on Ultrasoft in the May/June 1983 Softline. And thanks to Alan Dean Foster for taking the time to share his memories of the Shadowkeep project with me.)
October 18, 2013 at 12:58 pm
Very cool article once again… It is a gold mine
And I agree, you must be patient when playing Shadowkeep.
Don’t forget to turn the fast mode on of your Apple // emulator :)
October 18, 2013 at 6:10 pm
Indeed, writing the novelization of a game is no easy task. Shamus Young’s account of writing Free Radical based on System Shock is worth a read. (And so’s the book, but that’s besides the point.)
As for the game itself, it sounds like yet another case of overextending, ignoring hardware limitations and not prototyping first. At least they have the excuse of having done it in 1984…
October 21, 2013 at 5:03 pm
“some frustrated adventures” –> adventureRs?
October 22, 2013 at 6:04 am
October 21, 2013 at 5:41 pm
Ha, reading about what this game tried has been a kick.
For the last year I have been making a CRPG in Hugo. I’ve implemented all the non-text adventure stuff a CRPG needs, like making and saving characters, combat, graphics and a world model where there are six characters instead of one. I was describing this work in progress as being like “Bard’s Tale using Hugo.”
So you can imagine my surprise to learn about a CRPG from the 80s that tried to use a text parser in a role-playing game, haha!
Thanks for writing this up. I feel I can learn a lot from mistakes that Shadowkeep made. Some of this stuff I’ve learned on my own — for example, the bit where you have to hit return after moving. Jon Blask wrote some code to implement a “travel mode” in my game, so you can just move with the arrow keys.
(I also realized, thanks to this article, that I don’t have a way in my WIP to heal characters outside of combat. Slowly sneaking away to implement that right now….)
Thanks for exploring these games with less “fame” than the Infocom lineup, Jimmy. They are fascinating and it’s becoming very clear that there’s a lot to learn about them that’s directly applicable for games made today! =)
October 22, 2013 at 6:13 am
You’re welcome! Virtually all IF created over the last two decades has been very much modeled after the Infocom approach. For good reason mostly; I’ve certainly not found anyone in researching this blog whose games were so polished and smartly put together as Infocom’s, and I don’t really expect to. Still, these other traditions are worth exploring and maybe learning from.
Many of Shadowkeep’s failings come down to hardware limitations and a lack of usability testing. Conceptually, I find what it’s trying to do really interesting. So good luck! I’ll be interested to see the end result. At least Shadowkeep can serve up pretty good examples of most of the pitfalls you can fall into. :)
October 22, 2013 at 8:09 pm
Jimmy and Robb,
There’s one other 8-bit game I know of that tried this RPG/text-parser hybrid, Realms of Darkness for the Apple II, published by SSI in 1986. I own it and messed around with it while back – the graphics aren’t super-slow, thanks to it using the Graphics Magician technique seen in many graphic adventure games of the time, but I can’t remember how good/bad the interface is (I didn’t explore it that much). Might be worth looking into as another attempt at this hybrid.
October 23, 2013 at 8:31 am
Hmm, quite interesting. I wasn’t aware of that game at all. I may indeed have to look harder when the time comes…
October 23, 2013 at 10:19 pm
This does sort of feel like it’s getting a little late to reply (and also runs the risk of “shoehorning something only tangentially related in just because it’s something I can actually talk about), but the whole “press ENTER” aspect to this game did make me think of Dungeons of Daggorath, one of the TRS-80 Color Computer games that gets talked up today, and one where you had to type in two or three-word commands (although they could be abbreviated) and then press ENTER, with the clock ticking all the while… It could be a stretch to compare it to the numbers-quoting RPGs that have been described here, though.
October 24, 2013 at 5:31 am
I actually played Dungeons of Daggorath maybe ten years ago. It’s quite an achievement — but I don’t really consider it parser-driven, since the commands are all listed in the manual rather than “just type what you want to do.”
And no need to worry about commenting on older posts. I regularly get comments on things stretching all the way back to the beginning of this blog. I always get notified, and always read them with interest, as I’m sure do new readers just discovering those posts and old ones dedicated enough to sign up for the comments RSS feed. :)
October 22, 2013 at 8:00 am
Thank you for another exellent entry. When I saw the word “Ultrasoft,” my heart jumped for joy! (Actually, Ultrasoft was mentioned on the “Bookware” entry last month but it was so brief, just being a developer of Shadowkeep, and I expected more.) I’ve been interested in the company but no information available. Also I didn’t know why they stopped publishing their own titles themselves, and broderbund got to publish them instead. Now I got the answer.
And I’ve been reading your entries on telarium with much interest as well, especially Shadowkeep. The game was so tempting when I was informed of it for the first time (via some magazine advertisement, probably), but I cannot get to play it. Sorry to hear it’s so rare now. Thanks again.
Nori from Tokyo
August 24, 2014 at 10:53 pm
Greetings from your “neighbour country” Germany. I was very positively surprised when I found your detailed articles about many of the Telarium games with so much background info, this was tremendous fun to read. I have been collecting Telarium games for some time now, and I also managed to snatch a copy of Shadowkeep as well. I had always wondered why this particular game had not been released for the more promising platform of that time, the C64. I knew their lead platforms were the Apple II and the C64, the IBM PC and Atari ST came later. I knew there were people out there looking for a PC version of the game, but since the C64 version was already “missing in action” the chances of a PC version were next to zero. I think I also found a scan of an old product flyer with an offer for demo disks of Telarium programs, and Shadowkeep was listed as the only one that was Apple II exclusive. That confirmed everything. Because of your article now at last I know it was not just due to poor sales of the title, but due to the inability of porting the game to the C64.
I agree with you about the Telarium games in general: they had some of the most beautiful packagings ever with their gatefold design and colourful illustrations, definitely worth collecting.
I suppose the rather poor sales forced the company to sub-license the Telarium games to british company Audiogenic later on for a release in the UK. Would you perhaps know, if there are any MSX ports/versions out there, according to the Mobygames database a company named IdeaLogic might have released a port for the MSX in Spain?
That part with the sub-licensing still is not completely clear to me. Again thanks for the detailed insight into the history of the Telarium games, especially into this rare one.
August 25, 2014 at 5:17 am
I don’t have any specific details about the MSX Spanish translations other than the fact that they do indeed exist for at least some of the Telarium titles. Most American software publishers at this time weren’t terribly involved with translations for other markets, although they were happy to accept the money of companies in those markets who thought they saw potential and were willing to invest the time and money into making and marketing the port. A company called Starcraft became quite huge in Japan for a while porting American games to Japanese systems and occasionally cloning Japanese games for the American market. Various continental European companies had similar models. U.S. Gold published American games all over Europe. Sometimes they would translate the language for larger markets like Germany, sometimes just release it in English and hope for the best.
August 25, 2014 at 3:20 pm
thank you for quick reply. Yes, I do remember US Gold released several games under their “All American Games” label, for instance Ultima III or Masquerade (the old Phoenix/American Eagle) adventure and several SierraVenture titles. They also published the Beach Head Games and Raid over Moscow from Access Software and several Epyx titles. I just never knew why Audiogenic was chosen for publishing the Telarium games in the UK. It was probably one last effort to make some money from those Telarium titles.
Thank your for your help.
August 25, 2014 at 4:18 pm
For what it’s worth, my impression is that these arrangements were much more likely to be initiated by the company doing the “localized” version than the American publisher. So Audiogenic most likely came to Spinnaker. Why? It’s hard to say exactly. Maybe Perry Mason was particularly popular in Spain, in the same way that Germans went nuts for Dallas and Hogan’s Heroes. ;) Certainly text adventures in general were very popular there. Some years ago, when I was editing the old SPAG newsletter, I helped put together a history of Spanish interactive fiction. You may find that interesting: http://www.spagmag.org/archives/backissues/spag49.html.
November 20, 2016 at 6:56 pm
Only two years have passed since my last post, but I have an update on the Spanish Telarium titles. I indeed managed to locate a Spanish copy of Telarium’s Perry Mason, it is fully localised and called “El caso del asesinato en el Mandarin”. Published by Idealogic, distributed by Philips New Media Systems. Finally I have at least one copy for my own collection. I ahve not yet finished reading your history of Spanish IF, but will do so, when I have more time.
February 22, 2017 at 10:47 am
I think you’re being a little unfair to the parser on this and other games in your previous articles. Sure, they may have lots of flaws, but the idea of ignoring extraneous words was a means to allow the user to type in full sentences. It wasn’t misguided at all, just a bad implementation.
You say “who wants to type more than necessary under any circumstances,” which betrays your misunderstanding of the general player’s disposition at the time. As I remember, in reading contemporary magazines (and also playground chatter), the “holy grail” of a parser was one which could understand full English sentences; one on which you could type “TAKE THE SWORD AND GIVE IT TO ROB,” instead of the more terse “TAKE SWORD; GIVE ROB.”
One way to do this was to do full lexical analysis on the entire input string — a tall order indeed in a microcomputer of the times. Another way would be to ignore extraneous words such as articles, conjunctions, adverbs, and attempt to elucidate the verbs and nouns from the input. You see, the underlying engine still expects a VERB NOUN construct, but there’s a transformation step to extract those. That was the goal at least.
That Infocom managed to do this better than anyone else is no surprise, considering that building computer languages, compilers, operating systems — and performing research on artificial intelligence — was their actual forte in MIT. Everyone else who “got this” as the was forward was just trying to catch up with them, to various (and sometimes hilariously frustrating) degrees of success.
With all due respect, as much as I enjoy your writing (and I do sincerely, evinced by my continuing presence here, inching my way from beginning to end); sometimes I think your perception of these ancient games — a product of a more primitive and innocent time — is unduly coloured by a hardened and cynical eye which has been grated and jaded by decades of bad storytelling and game play mechanics. That’s absolutely fair to some degree: you are attempting to give us the unvarnished truth, warts and all, without the rosy-petaled glasses of nostalgia.
However, sometimes, it is just an exaggerated distortion of contemporaneous perceptions; and I have to wonder: Am I here to get a clear and honest view into a world long gone in order to understand it better, absorb its sensibilities, and compare it to the state of the art, as it were? Or am I here to get a critical review of a retro game with a slant towards my modern sensibilities? Sometimes I don’t know which one I’m getting.
Still, a most interesting read, sir.
March 13, 2018 at 2:39 pm
I think it’s perfectly clear from this and many preceding articles that Jimmy understands exactly what adventure-game developers were trying to do with their parsers in the early-to-mid eighties, and that he sees that goal as fundamentally misguided. And I have to agree. (I speak as one who in 1986 attempted to capture a subset of English in a YACC grammar for a game that I was writing. I was wise enough to abandon the attempt.)
The interesting thing in these games is solving puzzles. To solve puzzles, you need to be able to express what you want to do. That’s all the parser needs to provide: a way for players to express the solutions to the puzzles. 90% of the time, “SAW PLANK” is good enough. 90% of the rest of the time, one indirect object is enough: “GIVE WATER TO WITCH” (and even then, “GIVE WATER” when the witch is present really gets the job done). Only very rarely have I encountered a puzzle where more elaborate phrasing is required to convey the solution. There’s Sorcerer‘s “DRAW LINE FROM U TO V”, which takes a direct object and two differentiated indirect objects. Can you think of any more examples?
Peter R. Mason
April 20, 2019 at 7:30 am
WOW! I worked on the parser for this game. In fact, I was hired as a young lad by Ultrasoft as the “parser guy” in 1983. Alan Clark is much smarter than me (despite his lack of advanced degrees, unlike me), and he cleaned up my code many times. Pls don’t blame him. Blame me! But our fundamental parsing model was just wrong. These days it seems that natural language recognition is all about “big data.” Is “big data” the way that human brains work?? Your call.
As you correctly say, Shadowkeep’s parser, which I wrote, sucks. It’s too bad because there were a lot of really, really interesting ideas in the game. I went on to work elsewhere after Ultrasoft died. I’m sorry that people like Alan and Chris didn’t hit the huge game explosion later. Or maybe they did. Peter Mason
Peter R. Mason
April 20, 2019 at 7:54 am
And by “interesting ideas” in the game. I’m thinking “maze model” for example. That stuck in my head (ok I admit I have a short attention span). Every time I watch the Game of Thrones intro, I think of Michael, the game designer of Shadowkeep.
December 13, 2022 at 9:48 pm
Hey, Peter! This is Karri Barker (but now Loftus). Remember me? I worked with you up until the end of Ultrasoft. I have a lot of great memories of that time. The intro for Game of Thrones ranks as my favorite intro for any show ever, and now that you’ve mentioned it, I’ll think of Michael when I see it also. Those definitely were ‘the days’… Karri Loftus