Development Hell

“For Ultima IX we have two main goals: one is to build a truly epic conclusion to the trilogy of trilogies[...]in the sense of a great story[...]and the other is to build gameplay equal to or greater than our goals or the expectations of previous Ultima[...]And of course, while retaining some of the animations and details we've worked so hard for Ultima VIII, we're not building an arcade game rather than a role-playing game.”

Richard Garriott, MSN Q&A, October 22^nd, 1996

“For those who are worried about targeting enemies, you can press the A key, which makes the Avatar point directly at his opponent. Whatever your character does, in terms of movement etc., you will always be facing that creature and will be able to attack by clicking.”

Richard Garriott, interviewed by P. Stefan (Desslock) Janicki, Gamespot, 1998

“At this time, the 3D technology seen in Underworld cannot quite give you the holistic world view and interaction[...]that we enjoy in Ultima 7[...]The number of items to interact with must be kept low to keep performance high in a game like Underworld[...]but the gap closes[...]I am sure that not too far in the future, these two product lines will merge[...]in a completely virtual world in a game already in development.”

Richard Garriott, The “GAMERS' Forum's ULTIMA VII” Conference, July 8^th, 1992

Ultima VIII’s sales, though substantial, did not impress EA. The parent company had hoped for much more and thus further tightened their grip on Origin. EA management’s confidence in OSI hit an all-time low while rejections of new projects increased at an unprecedented rate. It wasn't just Richard who was in the crosshairs. Nearly half of the many projects launched failed to the point where development was cancelled and their teams were disbanded, often coinciding with layoffs.

EA's control over their subsidiary's activities had not just become firmer but, contrary to what had been promised at the acquisition, became downright intrusive. Suddenly, almost zero projects were given the green light as the high tide of funding and resources provided by EA had receded abruptly. Even worse, of the few projects that actually were greenlit, even fewer finished made it to production without being cancelled.

After openly investing so much, EA suddenly became obsessed with the bottom line and required cuts to meet increasingly tighter budgets. Often it fell to Richard to give the bad news to his colleagues. “Without question the hardest thing for me to do in business was, and still is, to fire somebody,” Richard laments. “In the early days of Origin, it had rarely been necessary. However, once we became part of EA and grew, then shrank very rapidly, it was much more common,”^[1]

Expectations for Pagan had been very high. When it became clear that Garriott's latest chapter of Ultima would not sync up with EA’s lofty prospects, despite the patch and the heartfelt message to his fans, other cuts were made. The team that worked on Ultima VIII had not stopped refining the product and improving it, even after the official launch in March and the patch immediately after. Towards the end of the year, a new and improved version of Pagan was slated to be published. This version, only on CD-ROM this time, would contain, among other things, portraits and voices for all the NPCs.

Expecting relatively low sales, EA cancelled the project, despite it being essentially completed and ready to go to press. The team had worked hard to fix most of the problems born of EA’s draconian December '93 deadline and the game was very close to what it should have been from the beginning. Given this, it's easy to imagine how hard it was to accept such a decision.

Another team was working on an expansion for Pagan, an add-on similar to Forge of Virtue and Silver Seed for Ultima VII. Like the above effort, this was very demanding work. Not only would this expansion add new content, but it was also supposed to restore content lost in the chaotic months before the original launch. The Lost Vale, as the infamous expansion to Ultima VIII was named, was supposed to give the public what the improved version on CD had failed to deliver, plus a long series of improvements, additions, and tweaks to the game engine.

The expansion’s team had been struggling for almost three months despite enjoying experienced staff members such as Sheri Graner Ray and John Watson. Watson, in a moment of levity, invented the “TinyTar,” via a spell that made the Avatar smaller. Everyone liked the idea so much that the team inserted it into the plot, requiring the player to perform the spell to enter an otherwise inaccessible area by passing through a small hole in the wall.

When the grueling work was completed, the team collectively let out a sigh of relief. Not only was the master disk ready for duplication, but the box and much of the packaging were ready for printing. It was at this time that EA cancelled The Lost Vale expansion.

Sheri Graner Ray explained in a 2013 interview, “It was directly due to a lack of sales. Back then you couldn't do anything as DLC, it had to have a box and go on a shelf with all the commensurate marketing dollars added to it. It came down to straight numbers. If the same percentage of U8 players bought the add-on as U7 and U7.5 players bought those add-ons, it simply didn't justify the cost of manufacturing and distributing the U8 add on.”^[2]

Statistics weren’t the only reason. In EA’s eyes, Pagan was now a closed chapter, an otherwise unpleasant failure that no enhanced edition or add-on could turn into something different. EA's attention was already on the future. Their eyes were now focused on Ultima IX, already with an uncertain future given the sales of Pagan and the unreliability of the releases in the series. Ultima IX’s uncertainty was only exacerbated by Richard’s proposal to bring Ultima to the internet and develop an online multiplayer version of the CRPG.

The Lost Vale was thus struck from existence, even though the master was ready and the add-on could fix much of what the players didn't like at all about Ultima VIII. After the project was scrapped, the game disappeared from the history of video games for many years, leaving behind only a few screenshots and the memories of the staff members who had developed it. Surprisingly, almost ten years later in September 2005, a box for The Lost Vale appeared for sale on eBay. It was a genuine copy of a prototype (though lacking a small part on one side), probably a box created before the publication, and therefore one of the rarest specimens in the history of video games. It was truly a unique piece, but unfortunately without the CD-ROM. Authenticated by Stephen Emond, who literally wrote the book on Ultima collecting, and Denis Loubet, who designed the artwork, it sold for almost $2,000. Later, in 2013, the same copy went back on sale, this time for €10,000. In 2018, a genuine production-ready copy of the box for The Lost Vale was retrieved by Enrico Ricciardi and then verified by Richard Garriott.

The mishandling and subsequent shortcomings of Pagan redesigned the relationship between EA and Origin Systems. With trust hopelessly lost, EA’s management closely monitored OSI’s activities. Moreover, EA's acquisition strategy, which had started with Distinctive Software and Origin Systems, had not stopped. Negotiations and market studies were underway for the acquisition of Bullfrog, the studio founded by Les Edgar and Peter Molyneux in 1987, that produced hits such as Populous, Syndicate, Magic Carpet, and Theme Park.

“At least partially as a consequence of that disappointment, management told me, basically, that they didn’t want me making big games like Ultima anymore. They wanted very small games and a lot of them. Instead of working on a $10-million game that would take years to make, they wanted games that could be produced in two months for $100,000, allowing them to spread their investment risk,” Garriott explains. It was a very strange strategy as it was in complete contrast to what EA’s top management was asking Warren Spector to do, which will be revealed later.

Ultima IX was certainly not a $100,000 project and, as a result, Richard's resources were siphoned off. Origin’s budget was always a bit burdened by Garriott's games but after the release of Wing Commander, the balance of funding had begun to tip away from the Ultima series. The eighth installment’s failure had only accelerated this shift of resources to other projects.

In 1995, Wing Commander IV: The Price of Freedom was at the height of its development. However, it wasn't the only game EA was counting on to get its subsidiary's accounts back on track. Starting with Ultima VIII’s engine, a team led by Tony Zurovec was working on a sci-fi action game named Crusader: No Remorse. Zurovec had joined Origin in early 1990s and had been part of Garriott's teams responsible for The Black Gate and Pagan. For Crusader, he held the dual role of director and lead programmer.

The project was both interesting and seemingly very promising. Pagan’s game engine, despite its obvious limitations due to rushed development, was designed to perform precise animations and action-based gameplay and not so much to handle the elements typical of role-playing games. And Crusader: No Remorse was precisely an action game. Crusader was set in a distant future where a single man, a former enforcer for the World Economic Consortium (an organization that had taken control of Earth's governments), had to infiltrate a secret base of the entity that he had just betrayed, destroy it, and take away the Consortium’s weapon of retaliation against the people of Earth.

Crusader was precisely the game to take full advantage of Ultima VIII’s cannibalized code and ultimately sold much better than its engine donor did. Origin, seizing this momentum, took the opportunity to put a sequel into production right away.

Garriott's first work on Ultima IX started from an engine very similar to Pagan's, likely an improved reissue of the final version the world never got to see. Given this, it's not surprising that in order to release Crusader: No Remorse within EA’s signature draconian timeframe, they started hijacking staff from Garriott's team and put them to work on Zurovec’s project. Zurovec who, by this point, had greatly improved the borrowed engine, allowing it to run at 640x480 resolution, over four times that of Ultima VIII.

In 1994, the first genuine stutter in Ultima IX’s development occurred when Garriott considered leaving Origin and EA. However, the situation was partially resolved when Richard unexpectedly managed to get the green light for his internet-based Ultima experiment.

The little progress made on Ultima IX at this point was essentially limited to the fruits of close collaboration with the Crusader team as the games were sharing an engine and even a common object database.

By the second half of 1994, the U9 dev team had been reduced to a skeleton crew. Under the direction of Mike McShaffry as lead programmer, there were only Denis Loubet, John Watson, and Jason Ely and Ely was soon moved to Crusader. This left the absolute minimum number of staff for the project to be considered still alive, though a more accurate term would have been “suspended animation.”

By this time, Richard had realized that he’d been maneuvered into a corner and that development on Ultima IX had completely stalled. He also posited that the use of Crusader's engine presented the risk of limiting his RPG and thus decided to abandon the parallel development Zurovec’s team, betting once again on new code for Ultima and forging ahead in their own branch. With the technical strategy decided, Garriott invited an old childhood friend of his to help turn the project around and resurrect development. Robert White, the dungeon master specializing in demonic, dark, and brutal adventures was thus brought into the fold and given the title of associate designer.

Robert, brand new to the video games industry, joined Ultima IX’s team when the project was ostensibly still in its infancy, even though it had already started some time ago. He recalls, “Richard and the two senior designers had sketched out a ‘point A to Point B’ kind of plot. They knew how they wanted to start and where they needed to be at the end and had some interactions figured out.”

Richard intended for Ultima IX to be the closing of his trilogy of trilogies, a grandiose, hopefully epic chapter, with an ending that would be the capstone to one of the most important series in the history of video games. For Garriott, Ultima was much more than a computer game or a simple commodity that brought him his success. He explains, “In Ultima IX, it’s the end of the Guardian, not only that but it’s also the end of this whole Avatar business we’ve been going on about, so Ultima IX is actually the conclusion of Ultimas I through to IX. Ultima X will be a completely fresh start and probably won’t even be called Ultima X.” Additionally, at this stage of development, Garriott was convinced that the complete annihilation of Britannia would be the appropriate ending for his series!

At this phase of development, the plan was to return to the party system of games past. Thus it fell to White, who was tasked to develop an epic plot that started with the Avatar’s escape from Pagan, involving most, if not all, of the famous companions of virtue, and culminated in the wholesale destruction of Britannia. This would have been challenging at the best of times, but as we know by now, development under EA’s ever watchful eye was far from an ideal situation.

Struggling against EA's increasingly manipulative micromanagement, Garriott managed to channel new energy into the project. Robert White again, “Richard had also drawn upon several of the really powerful artists available at Origin to start doing concept art and to work with the designers on a look and feel[...]from the interface to the in game movie scenes.”

A special tool was created to allow writers to write both dialog and their in-game ramifications with the ability to reference and modify quest flags, items in the party’s inventory, and the Avatar's companions. The tool’s purpose was to allow writers and designers to work on the plot before the engine was finished. White continues, “We four members of the design team (John Watson, Brian Martin, Chuck Zoch, and myself, and Richard of course) came to an interesting early agreement. With the storyline and deep interaction required of a great Ultima game, we thought we should be able to play the entire game, front to back, by conversation only. A new and extraordinarily talented programmer had come on board (Ragnar Schuermann) and his first task was to build us a conversation engine.”

Once this novel engine was complete, the designer team set to work on the conversations. White, meanwhile, had been working on the plot of the game for a while by this point. A few years later, a version would be published online under the name "Bob White Plot," although it was a collaboratively written document. White clarifies, “That document, the one that is circulating online, called the ‘Bob White Plot’ is one of the last copies of my work with Richard, Chuck, Brian, and John.”

Shortly after the release of Ultima IX, in response to strong initial criticism, Robert wrote a post on the internet explaining the troubles with development, the changes to the game after his departure from Origin, and revealing a summary of his forgotten plot for the first time in 1999. The full version was made public only much later, in 2008, when EA Mythic, heir to OSI, had entrusted some enthusiastic fans with the task of recovering, cataloguing, and storing large amounts of material that had been Origin’s.

With the lackluster response to Ultima VIII and its ending, in which some fans were not sure that the world the Avatar returned to was even Britannia, Richard promised a return to the familiar world where most of the previous episodes had taken place. Thus, ignoring the ending of Pagan, the Bob White Plot began with the arrival of the Avatar on the top of a mountain in Terfin, near the Guardian's fortress, where the player's alter ego was attacked immediately by a Wyrmguard.

Miraculously escaping death via intervention by the Time Lord, a.k.a. Hawkwind the Seer, who had previously helped the stranger in Exodus and the Quest of the Avatar, the player was transported to Stonegate and brought up to speed on what had happened to Britannia during their absence. Giant columns had risen throughout the kingdom, the land had become infertile, and a dangerous disease was spreading. Initially, the cities had worked together to cope with the crisis: the magicians of Moonglow magically produced food and shared it with the other cities while the monks of Britain, having discovered that the poison of the silver serpent could be distilled into a palliative for the disease, had made this cure available to the denizens of the other towns.

Lord British had also fallen ill. Feeling his strength failing, he formed the Tribunal, a committee comprising the leaders of the major cities and had given them the task of leading the kingdom. Unfortunately, several disagreements soon arose within the Tribunal and the cities stopped cooperating, ultimately preparing for war in order to compete for the limited resources available.

When the Avatar arrived, the Guardian had begun inserting images into Lord British's dreams, revealing to the ailing monarch the events that had transpired in the world of Pagan in an attempt to discredit the paragon of virtue. The looming civil war and the deteriorating relationship between Lord British and the Avatar were the central themes to much of White's plot.

Only the capture of Lord Blackthorn, the Guardian's ally throughout the game, allowed the Avatar to regain Lord British’s confidence and reunite them before the final clash. It was at this point, however, that the player discovered that the civil war had been a tool to distract the Avatar. The Guardian’s true intention was to accumulate enough energy in the giant columns that it could be explosively released, resulting in the destruction of Britannia.

Lord British, having no choice but to warn everyone and prepare them for the impending catastrophe, asked his subjects to gather at Skara Brae, where they would be protected by the runes of virtue. Once the populace was safe, Lord British and the Avatar went to confront the Guardian directly at Stonegate and defeat him, despite Lord British being wounded (in a cinematic sequence that unfortunately no longer exists). Hawkwind appeared at this time to warn that even if the Guardian is killed, he would come back to life through the energy released by the columns and using the Armageddon spell was the only way to defeat him forever, even if it would destroy Britannia in the process.

Lord British, being injured, was unable to perform the apocalyptic incantation alone and the Avatar decided to stay and assist, sacrificing himself in the process. The Armageddon spell tore the whole world apart, pulverizing it into a cloud of fire. The only inhabitants of Britannia who could save themselves were those who had listened to Lord British and fled to Skara Brae. The city literally fell off the ground just before the shockwave hit it in a scene reminiscent (and even mentioned explicitly in the plot) of the cover of the album Yessongs by the British band, Yes.

It was now 1996, a critical year of development for Ultima IX and the fatal one for Origin. While Richard was using all his influence to hold the ninth episode together, OSI was beginning to crumble. Despite everything, Richard somehow managed to cobble enough resources together to begin creating the expensive animated sequences and start recording audio for voice-over.

The 3D animated sequences were drawn according to the Bob White Plot: the arrival of the Avatar at Stonegate with consequent attack by a dragon; the conversation between the Guardian and Lord Blackthorn; the dreams of Lord British, nefariously manipulated by the Guardian to make the king believe that his protégé was out of control; the execution of the Armageddon spell and the subsequent destruction of Britannia.

With Ragnar Schuermann’s conversation engine, the writers had been working for months crafting copious amounts of dialog and testing how the game worked, all without having a working game in hand. The text was ready and the recording sessions were started, though this time not directly at Origin, but rather at the prestigious Magnolia Sound Studios in Hollywood.

In the September 20^th, 1996 edition of Point of Origin, the then-producer of Ultima IX, Joye Price (McBurnett at the time), commented on the recordings, “All in all, the session went well. Stu Rosen–the director–was amazing. We all had a blast.” Next to Price's commentary, you could see the portraits of the voice talent, all professional actors, who were chosen for the cast of Ultima IX. Keith Sjarabajka, at the time known for his part in Clint Eastwood's film A Perfect World, as the Avatar. Clive Revell, voice actor of Emperor Palpatine in The Empire Strikes Back, as Iolo. Kay E. Kuter, from famous TV shows like Green Acres and Petticoat Junction, as Lord British. Also present were Scott Cleverdon as Lord Blackthorn, Cree Summer as Raven, Scott Bullock as Samhayne, and Maurice Lamarche as Shamino.

Not many details are known about this stage of development. Some information was leaked in the specialist magazines of the era but few pictures were made public. The graphics engine had been rewritten and, though continuing to maintain an isometric perspective, it was equipped with software rendered 3D polygons. Between 1995 and 1996, the early pioneers of 3D graphics chips, including nVidia and 3dfx, began showing and commercializing the first graphics accelerators. Prior to 3D hardware acceleration, three dimensional graphics were rendered by the software using the main CPU, which greatly limited the resolution and number of polygons that could be used.

Equipped with a mobile camera rotating around the vertical axis that could zoom in and out, the new graphics engine was somewhat reminiscent of Syndicate Wars (1996), one of Bullfrog’s first games as a subsidiary of EA. The game map would be much larger than Pagan's, at least four times as much, occupying 265x265 screens, while the game system had to take a step back in time, giving the player control over an entire party again, rather than just the Avatar.

Soon, however, the first problems with the party system arose, mainly due to game performance related to the previously mentioned software rendering. The developers realised that the game tended to slow down unacceptably when there were a large number of characters on the screen.

One of Herman Miller's first assignments, now a veteran of the series, was to try to improve the code and the game’s performance. He recalls,

 “The initial development for Ultima IX was done with graphics rendered in software, since 3D graphics accelerator cards were just starting to come out for PCs and weren't yet widely adopted for games. Like the earlier games, the world was viewed from above, but internally the world was actually stored in 3D coordinates instead of just being shapes tiled on a 2D grid. I remember that the frame rate was slow, and I did some profiling to find that most of the time was being taken up just copying graphics data around in memory. By rearranging some bits in the data format, I was able to cut the buffer size in half, which greatly sped up the frame rate, but it was still slow.”

Despite these challenges, the game was rather far along in its production cycle. Audio recordings were made, movies were ready, texts were written, and the design team, starting with conversations and interactions with NPCs, was rapidly crafting areas of the world, populating them with characters, and writing quests. Robert White again, “Richard insisted (quite rightly) that the sub plots could not be just singular retrieval games, but that in the end each one of them would advance the plot [and] provide clues along with any rewards. No sub-plots for just their own sake.”

Also around this time, rumors about the game began to leak. Previews appeared in some magazines for the Sony Playstation console, suggesting how possible, even probable, a console version of Ultima IX was. Had that been the case, it would have been the first title in the series that was developed cross-platform for PC and a console simultaneously. Porting to consoles had happened in the past, for sure, but they were always after-the-fact efforts that were not in tandem with the original development.

Returning to Ultima IX’s ill-fated development, the game had already survived having its team be cannibalized by the Crusader team. However, an expected new obstacle appeared on the horizon. This time it was Ultima Online's turn to inflict the fatal blow to the second phase of the ninth episode’s development.

Richard had struggled to make EA's management understand that his new multiplayer experiment could be revolutionary and, at the same time, wildly successful. When he finally convinced them, EA's directors moved all available resources to Ultima Online and the first victim of the reorganization was Ultima IX, which was put in suspended animation for a second time. Robert White laments,

“The move to UO was not unexpected, but I was puzzled since EA liked to drop games that fell off the radar. EA already had a vested investment in UIX so delaying the game might keep the game costs reduced in the meantime, but keeping EA wanting to produce the game would be difficult. EA was kind of fickle and would change their minds at the drop of a hat, but they always wanted a money maker. Richard could see the changes coming in the gaming world about mass multi-user games and some of us could see what was happening although the first ones we looked at were more like glorified chat rooms. We had a feeling it would be big, but the concept was complicated.”

Miller was also moved to the UO team, though briefly. He explains, “It became apparent that Ultima Online was going to need a larger team, so Ultima IX was put on hold and everyone on the Ultima IX team was moved over to Ultima Online. I was only on Ultima Online briefly for a month or two, doing optimizations to make the game faster.”

The mass migration of Ultima IX developers to the new game was not altogether welcomed. In addition to irritating many members of the redirected team for having to suddenly change directions completely, they also sensed that the break would have serious consequences on the single player game’s production, perhaps even becoming a permanent cancellation. Because of this, the two teams did not fully integrate.

The continuous delays on Ultima IX also messed up plans for a special project that aimed to republish all the games in the series in a single special edition on CD-ROM, along with the final chapter. The repeated postponement of the final episode had put both Origin’s top management, and that of EA, in the difficult situation of needing to make progress on the special project even though the final chapter was not yet complete.

Corey Roth, a young student with a passion for Ultima, tried to get in touch with Garriott to ask if a port of Akalabeth was available on the PC platform. Of course, Roth had no way knowing about the upcoming special project yet. Roth explains, “The answer I got back actually surprised me. He said ‘No, there wasn't a port and if you could write one in good C++ code, we could look at including it in the upcoming release of Ultima IX.’ As a total fanboy, I was overjoyed at the opportunity, only one problem. As a 19-year old aspiring computer science major, I barely knew any C++ at all. I had to learn it overnight.”

Roth set to work enthusiastically. He rapidly learned C++ and printed out Akalabeth’s BASIC source code on thirteen sheets of paper and set himself to programming immediately. One of his biggest obstacles, as many developers of the first era of microcomputers had already experienced, was the abstruse handling of high resolution graphics on the Apple ][. Having overcome this first major hurdle, Roth, who had begun using Turbo C++ 3.0 as a compiler, decided to switch to DJGPP and the Allegro library in an attempt to fix some graphical glitches he had come across.

The work was completed over a handful of months in the student's free time. To congratulate and thank him, Richard invited Roth and Nathan Tucker, the friend who had helped test the software, to the Origin offices for a picnic. After the meal, Richard gave the duo several gadgets and games from OSI and EA’s catalog. Roth recalls the initial intended approach of the special project, “The original thinking is that you would be going through Ultima IX and at some point run into the ‘Book of Ultima.’ From there you could launch into any of the previous Ultima titles including Akalabeth[...]Ultima IX was hugely delayed and that never happened. EA did decide to release the Ultima Collection though and [Akalabeth] was included on that CD and of course now you can find it for free on GOG.”

Even though all the staff had been shifted to Ultima Online, not everyone had resigned themselves to completely abandon Ultima IX. Mike McShaffry, for example, hadn't given up and continued to work on the ill-fated project in his spare time. He experimented with new technologies in the freedom allowed to him by virtue of him using his personal time at home in the evening. Among the various experiments he did, he started working with the first 3D libraries for the new graphics accelerators that were appearing on the PC market.

One of the first companies to dabble in the 3D hardware market was 3dfx, founded in California in 1994 by Gordon Campell, Gary Tarolli, Scott Sellers, and Ross Smith, all former Silicon Graphics engineers. By 1996, 3dfx had entered the graphics card market, offering the first 3D graphics accelerators as high-end, high-cost products, extremely powerful and far superior to the other cards currently on the market: the Matrox Mystique, S3 ViRGE, PowerVR Series 1, and ATI 3D Rage.

Working with his Voodoo Graphics PCI card, McShaffry tinkered with the existing codebase, adding a 3dfx chip-based rendering. Satisfied with the result, he showed it to his colleagues.

Here versions of the story differ in some ways. McShaffry himself stated in an interview that he ported the Ultima IX graphics engine using the proprietary 3dfx API, Glide, to take advantage of the hardware graphics acceleration provided by his Voodoo video card.^[3] According to Bill Randolph, however, McShaffry had only managed to implement software rendering during his experiments at home.^[4] Mike recalls,

“Converting Ultima 9 to Glide wasn't really all that hard, and it wasn't a big change to the engine. When everyone saw the results, I think everyone was really happy, even Richard. They had mostly given up on Ultima 9 at that point, and my work gave it a second chance. Ultima 9 was a 3D engine originally, but with all software rasterization, which was simply too slow for the polygon density we wanted. I wasn't on the Ultima 9 team because by the time the project was started again after Ultima Online, I had already left the company.”

McShaffry's work was impressive and stirred things up, convincing EA’s frustratingly indecisive management to defrost the project once again and change the intended direction of the game. Specifically, it was decided to take advantage of the hardware graphics acceleration provided by 3dfx video cards, harnessing its full potential via the Glide library.

It was a sensible choice as 3dfx was the leading company at the time. Another company, Nvidia, had actually created the market initially by launching the first 3D chip, the NV1. For a number of technical reasons, however, it wasn’t well received by developers as they realized that these first chips were substantially slower at rendering 3D polygons than even the software rendering method.

3dfx, on the other hand, introduced the first successful 3D chip, convincing many developers to implement support for the proprietary Glide API in their games. Games of the era often gave users the between software rendering, possible on any platform as long as it was powerful enough, or hardware rendering with Glide, possible only with compatible cards, but with much higher quality.

One of the earliest games to competently leverage Glide to the amazement of gamers (and surely to tremendous market success) was Unreal in 1998. However, Origin had already implemented the 3dfx API into a game that was published a year prior. Wing Commander: Prophecy, the final chapter in the epic space based saga, had come out without the support of its original creator as Chris Roberts was already working on Starlancer.

Though McShaffry’s work rekindled EA’s interest in Ultima IX, he curiously remained on the Ultima Online team and by the time Ultima IX's development machine was up and running again, McShaffry had already left the company. According to Bill Randolph, the removal from the Ultima IX team had utterly broken the lead programmer's heart. After all, Mike had poured copious amounts of unpaid blood, sweat, and tears into keeping Ultima IX from dying so it’s understandable how devastating it must have been to be yanked away from the project. Years later, McShaffry wrote a chapter entitled “I never gave up on Ultima IX” in his book, The Secret of Game Coding: Game Coding Complete Guide.

Meanwhile, another important event occurred, the consequences of which no one at Origin could have anticipated. EA had bought Virgin Interactive–which in turn came with Virgin’s recently merged Westwood Studios–and moved two former Virgin executives to Origin: Neil Young^[5] and Chris Yates. The former would become vice president and general manager, while the latter would take over as vice president and chief technology officer. It was a move designed to improve the parent company's control over the subsidiary, as Origin quickly discovered.

According to sources preferring to remain anonymous, between 1997 and 1998, Richard Garriott's position, despite the success of Ultima Online, had become precarious due to the fact that he was considered a “liability” as his series was now dead and/or otherwise out of fashion. The only thing missing in EA’s eyes was a compelling reason to remove him from the production process and oust him from Origin Systems. Again, according to these sources, a possible failure with Ultima IX would have been the tantalizingly ideal excuse to remove the founder of Origin from EA’s subsidiary.

To help relaunch U9’s development, staff reinforcements arrived in the form of two former Westwood employees: programmer Bill Randolph and producer Ed Del Castillo. By 1997, Westwood had merged with Virgin Interactive, so the two former Westwood folks had already worked with Young and Yates, which would end up being relevant.

Del Castillo first met Richard Garriott in 1997 on one of his business trips to Texas when he was working as an outside producer for Origin while collaborating with Firaxis. Richard wanted to give a new direction to Ultima IX’s development and Del Castillo, who had always been a big fan of the series, offered to help out. The situation he found himself in, however, was not at all what he expected.

The team, weakened by the reallocations, first to Crusader and then to Ultima Online, was frankly under-powered, especially given the ambitious goals of the project. Origin’s legendary artistic department, easily one of the most talented departments in the video games industry at the time, had done incredible work so far, producing rich and impressive environments. Unfortunately, the programming aspects of production suffered from serious shortages of staff and, above all, a lack of organization and work planning.

Ultima IX’s development started in 1993 from the code written for Ultima VIII. Over the years, several programmers took turns modifying what was already there as well as adding new features. The 3D engine was not natively implemented, but rather was the result of a series of readjustments and subsequent transformations. Generations of programmers, many of whom were no longer on the team–and some no longer even in service at Origin–had taken turns, working literally in layers on the work of their predecessors.

There was so much code, often quite outdated, and so many features, that no one really had any idea how many elements had been implemented and then abandoned over the years. Bill Randolph explains “There was a lot of tech in there, a lot of very old tech. So it took me a while to ramp up on that, and start working on scheduling things, and trying to figure out how to move this forward. That took the majority of my time.”

The amount of ideas that had been developed, implemented, and ultimately abandoned without the related code being removed was such that there were always surprises in store. And bugs, of course. At some point during development, it was decided to implement naval battles in which the vessels could be damaged to the point of sinking. The design had changed, of course, and the battles were removed, however, the residual code had not. So much later, when someone designed a quest that consisted of landing on a monster-infested island, the evil mob recognized the seafaring vessel as a possible target and attacked it until it was “dead.” The ship promptly sank.

The game's engine also performed at only a handful of frames per second, so it was unplayable, even on the most powerful computers available at the time. It was later discovered that the slowdowns were due to the fact that, under the many sedimentary layers of code, there was an entire duplicate combat system, probably inherited from previous versions of the game, which consumed valuable system resources without any benefit.

Of course, at the base of it all, though thoroughly modified, adapted, expanded, and transformed, was the code written for Crusader, which itself was an appropriation of Pagan's architecture. “The engine was still the Crusader/Ultima 8 code,” Bill Randolph notes, “it had just been iteratively worked on to become 3D.”

At this time, Richard was quite busy with other business and personal travel. Additionally, Ultima Online’s development had reached a critical stage and if that wasn't enough, Richard was routinely occupied with building his new home, Britannia Manor Mk. III. He objectively had too many commitments and, despite his best efforts, was struggling to follow Ultima IX’s development as closely as was needed. Bill Randolph explains, “There were times that we were hurting for more of his involvement. And, you know, he had a lot of other obligations, and he had a lot going on, and a lot of other interests that he was pursuing too. We could not have Richard as much as we wanted.”

Another serious problem for the final episode’s development was the lack of communication and clear leadership. After all, the team had been dismantled and reconstituted several times by this point and Del Castillo's arrival only complicated the situation. Being the last to arrive, he was not readily accepted by the group. As Ultima IX’s development team had been transferred little by little to Ultima Online, the tension increased to match, setting the tone for how UO’s development would play out. This, however, will be discussed more later.

With tensions skyrocketing, arguably the biggest concerns for Origin's staff was whether or not they would lose their jobs. 1996 had been a dramatic and difficult year. It was marked by several rounds of layoffs, including many experienced employees. Several of those that received the axe were genuine industry veterans with quite a few successful published titles under their belt. There was a strong sentiment that Ultima IX could be nuked on a whim at any moment with its staff relegated to other projects or summarily terminated. The latter possibility, besides being the worst, was also the most likely, unfortunately.

Taking a look at the convoluted state of the game, Del Castillo proposed to start with a clean slate and begin development from scratch, especially with regard to programming. The goal was to create a native 3D graphics engine with only the features needed for that particular game, unburdened by the vestiges of leftover code. His request was met with a resounding “no.” Undeterred, Del Castillo commenced work using what was available to him and his first task was to convince Richard to change the plot of Ultima IX.

Feeling that the apocalyptic ending could potentially be indigestible to a large segment of long term fans, Del Castillo proposed a different plot. Some points were left purposefully in common with Bob White’s version, which would save as much work already done as possible.

Del Castillo’s reworked plot began with the corruption of the “towers of virtue” which were transformed into “towers of vice” by the Guardian. Thus corrupted, the towers began the slow destruction of the world. Via the purifying intervention of the Avatar, they returned to their native state and the danger was diffused. The first encounter with the Guardian followed. This is where the player learned that the Avatar and their nemesis were two sides of the same coin: the good and the evil side of the Stranger's soul, torn into two parts when he entered Britannia.

Realizing that the only way to truly defeat the Guardian was to reunite the disentangled halves of the Avatar’s soul, the player had to sacrifice themself. Upon reunification into one being, the Avatar/Guardian disappeared from the world of Britannia, which was now both purified and restored to its original condition, thanks to the player’s sacrifice.

The finale, in a nod to long term fans, would end in the same living room where Keith Berdak's painting could be seen hanging next to an entertainment center sporting a CRT television. This final screen, besides harkening back to the end of the second trilogy, would have had the same graphical look, including the 256-color palette of yore. With the player poetically returning to the living room of Ultima VI, they presumably lived the rest of their years without ever being summoned again to Britannia. Or perhaps it was to suggest that the entire Age of Armageddon was all just a daydream in the comfy chair facing the television. What do you think the forgotten ending means?

The revision of the Bob White Plot was therefore not a mere simplification. Rather, it was motivated by Del Castillo’s desire to make improvements and, above all, to create a plot that would not culminate with the destruction of Britannia. His solution ended with the Avatar’s ascension into the ethereal void thus closing a multi-game narrative circle. This was the moment that the product so far advertised only as “Ultima IX” definitely became Ultima: Ascension.

Another justification for this major revision was the perceived need to both normalize the storyline and remove the remnants of subplots littered throughout by the generations of designers who had worked on the project during the years of “development hell.” For example, there were characters who had been introduced by designers that were subsequently relocated to other projects (some permanently) or weren’t even employed by Origin any longer.

The party system, which already seemed difficult to implement in the previous version, was completely removed. It was initially proposed to replace the party system with the ability to selectively control other characters, such as Lord British or Raven, in specific sections of the game. This possibility was discarded as well because it was also deemed too difficult to implement. EA was already eager for the game to be finished soon, but even further misadventures were on the horizon and production would have to be stopped yet again.

Without a start from scratch, development continued to encounter major obstacles. Del Castillo had managed to wrangle Seth Mendelsohn, a new designer, on board to help the anemically small team. The programming side, however, was substantially undersized compared to similar projects. At this point, EA’s top management tightened their belts once again, and were attempting to make as efficient use of their resources as possible. Paradoxically, expectations were still high and a deadline was virtually impossible to set. Additionally, sudden unreasonable requests, such as a late stage implementation of multiplayer, further disrupted the timeline. Del Castillo strenuously opposed the marketing gimmick of multiplayer, thankfully winning out over the execs and maintaining the episode’s single player status.

Project support from the increasingly tight-fisted top management had decidedly been reduced. That, coupled with increasingly contradictory demands, was a clear portent of EA’s strategic retreat that would take place in just a few years, culminating in the subsequent closure of Origin Systems.

According to internal OSI sources, the lack of support was deliberate. If Ultima IX managed to be finished in time but failed commercially, it would have been the final mistake in EA’s eyes for Garriott, who had invested the remainder of his clout in the project out of love. If Ultima IX had not been completed in the allotted time, however, Richard would still have been an easy target with ultimately the same end result.

The political situation at Origin was complicated in the latter part of the company's life. Bill Randolph recounts, “[Del Castillo] can be viewed as someone who is vilified a lot, I think, and I don’t really think that’s fair. Ed made some enemies while he was at Origin; he didn’t pull punches, and he didn’t play political games. He was just a ‘no bullshit’ kind of guy.” When the marketing department decided that Ultima IX should have a multiplayer mode, which Richard had rejected only after initially genuinely considering including it in his game, Del Castillo had no problem saying clearly that it was not feasible up front. Not only that, he flat out stated that it was a bad idea.

Between the ever-present fear of losing their jobs and the tangible disappointment at EA's lack of support, the team’s morale sank. Del Castillo's work was seen as an intrusion by an outsider and the producer himself found himself between the hammer and the anvil: from above, there was a dearth of support to competently continue development and from below, there was ever growing resentment.

Prior to the redirection of the team for UO, the game was otherwise well underway and it would have been an Ultima episode much more in line with the previous games than the one that ultimately saw the light. Though still in 3D, the second version of Ultima IX intended to have a party system, subplots linked to the Avatar's companions, and the familiar top-down view. The hijacking of the development team and the decision to revolutionize the engine by making it immersively 3D with a third person perspective had completely changed the project and many designers and developers felt understandably disappointed that their work would not see publication.

The generally agreed upon theme for this period of development is “a frustrating lack of communication.” The perception was that Del Castillo was demolishing or at least oversimplifying what had been built with malicious intent. Ed’s background at Virgin and his connection with Neil Young and Chris Yates (the two men EA inserted at the top of Origin), combined with all of the previously mentioned development woes, caused the team to remain firm in their belief that the producer was a just foreign body who was not interested in creating ties. "He also didn't build bridges; he wasn't a bridge-building guy at that time,” Bill Randolph notes.

The objective of this text is not to decide which version of Ultima IX is best. However, from all of the information collected while researching this book, it's clear that Del Castillo was trying to create an RPG that lived up to the high expectations of the hardcore fans of the series as opposed to a simplified and mainly action-oriented one. Evidently, even his collaborators were strongly committed to this goal but it was impossible to reach as everyone’s efforts lacked a unified and focused direction. Additionally, executive direction was sorely lacking, which continually impeded the creative and production processes.

The development team, convinced that a start from scratch could have straightened the project, did not know that Del Castillo had carried out the request already and that it had been rejected. Randolph recalls, “We should have started from scratch. Absolutely. There was never a time on the project where we had the political clout to tell the company: ‘You know what? We’re gonna start clean.’ The project had been through multiple iterations, and even though I wasn’t there for those, the company would never have gone for a complete fresh start with the code; that would have been viewed as a complete disaster.” With the inability to start fresh, production had to continue with what was available which meant trying to improve the performance of the engine and remove at least part of the vestigial code left from previous stages of development.

In February 1998, the first screens of the new third-person 3D graphics were shown to the press, but within a couple of months two designers, Marshall Andrews and Dan Rubenfield, left EA, accusing the company of wanting to sink the project again.

Despite hemorrhaging colleagues, Del Castillo held fast and prepared a teaser trailer for E3 1998 with several drops of action and combat, frenetic editing, and a heavy metal soundtrack. Part of the fan community was already in turmoil due to the material already made public and now an online controversy arose. Del Castillo felt the need to respond personally, “Ultimas are not about stick men and baking bread (even though I saw a flame that said that is exactly what Ultimas were about) Ultimas are about using the computer as a tool to enhance the fantasy experience. To take away the clumsy dice, slow charts, and paper and give you wonderful gameplay instead. They were never meant to mimic paper RPG's, they were meant to exceed them.”^[6]

It wasn't the first time a rift had appeared between Ultima and its fanbase. Richard had repeatedly renewed his series over the course of nearly two decades and eight episodes; previous innovations had already alienated parts of the audience at different times. Continually experimenting, Garriott had completely changed the formula of his games more than once: starting from RPGs with punitive and hardcore mechanics, which forced the player to survive increasingly complex and protracted battles, proceeding through an era of unbounded interactivity and exploration, and arriving at games where the narrative had assumed the most important role. At each step there was a part of the fan population that didn't like the change and reacted by writing letters that Richard still keeps to this day.

With the advent of the internet, however, something had changed for the worse. The discussion over the trailer, judged to be too much action and not enough Ultima, became rather heated, especially when Del Castillo stated that "Ultima is not about baking bread." Of course, the controversy was cobbled together from evidence comprising only this trailer for a game that had been ostensibly in development since 1993, various screenshots, all with different graphics engines, and previews introducing features and mechanics that were long since abandoned or heavily modified since then. It’s reasonable to assert that any feelings of betrayal could (and possibly should) be attributed to fans having unsubstantiated expectations blown out of proportion, but such is the way of the internet.

The Usenet discussion groups were filled with open letters, some including full on accusations, to Origin’s management. The main target was Del Castillo, as he was perceived to be responsible for the shift to action-oriented gameplay and loss of Ultima's identity in the upcoming chapter. The mudslinging machine was in action and operating at full speed. As you'll see later, within a few years even Garriott himself ended up in the sights of the worst flamers.

Eight days after Del Castillo's response to one of his fans, Robert White announced that he was going to leave Origin Systems to join Andrews and Rubenfield, two of his former colleagues, at IonStorm, the company founded by John Romero. Doom's co-creator was busy with Daikatana, another game that would remain in its own development hell for a long time. Nevertheless, Romero’s company was about to open a studio in Austin and a team led by Warren Spector would soon get to work on the famous and successful Deus Ex, a game based on concepts that EA had discarded.

Less than a month later, Del Castillo also left Origin. The official press release only mentioned “philosophical differences,” announced that Richard Garriott himself would take the lead on the project, and reassured the public that Ultima IX would not be delayed or cancelled.

Having departed Origin Systems, Del Castillo founded his own company, Liquid Entertainment, where he went on to produce several real-time strategy games such as Battle Realms, Lord of the Rings: War of the Ring, and Dungeons & Dragons: Dragonshard.

With the departure of Del Castillo, the atmosphere was both dramatic and chaotic. The only surviving designer, Seth Mendelsohn, had been promoted to lead designer and Garriott himself had taken the role of director for the first time since Ultima VII. The two immediately set to work to restore some order and, of course, to decide how to start again.

The plot was again reworked. Bob White's writing was centered on the civil war and destruction of Britannia, while Del Castillo's rewriting was centered on the purification of the towers and the ascension of the Avatar. Garriott and Mendelssohn decided to return to a more classical plot, centered on the virtues: the Guardian had stolen the runes of virtue and corrupted them; it was up to the Avatar to find the runes, which had become corrupted glyphs at the base of each column, and purify them. Much of the game, therefore, was reminiscent of Ultima IV as the player had to travel around Britannia, recover the glyphs, and meditate in the shrines of virtues to purify them and deprive the columns of their destructive power.

The ending remained unchanged, however, with the Avatar discovering that they were the good half of a bifurcated soul and then sacrificing themself with the Armageddon spell to become one with the Guardian and ascend. However, unlike how Del Castillo had conceived it, the finale did not culminate with a return to the pizza box littered living room of Ultima VI, but rather with a starry sky and the outline of the ankh, the symbol of Avatarhood and the virtue of spirituality.

Only a few weeks after Del Castillo left, the remaining staff began to rework the engine and heavily modify it for the new draft, the fourth one since development had commenced on Ultima IX.

The moment could not have been less propitious, unfortunately, because just a few weeks later, EA's managers asked for a demonstration of the engine's functionality and the poor performance almost convinced them to close the project forever and trash everything. Only Richard’s passionate intervention, at least for the moment, saved the game. However, time was no longer a resource that could be trusted for EA to give. Richard put all his credibility on the table and the final showdown was drawing near.

The new deadline, this time final and not extendable, was Christmas ‘99. Perhaps hoping that the project would collapse in on itself and therefore avoiding the embarrassment of publishing a game at the time seemed to be utterly hopeless, EA began to cut resources again. This time, diverting them to Ultima Online 2. Though curiously, UO2 would ultimately never see the light of day.

Faced with a looming deadline and limited by scant resources that may as well have been doled out via eyedropper, Garriott made the difficult choice to cut everything that could possibly be taken away. The first cut decimated NPC functionality, summarily stripping them of their ability to perform daily actions, an important feature present for years in many chapters of the series. Even the possibility of following the Avatar in some missions, a mechanic introduced to compensate for the total lack of a party, was culled. The biggest hurdle was that the staff hadn't been able to implement an effective and reliable system of moving NPCs in the 3D environment and the characters tended to get stuck in front of invisible obstacles, sink into the ground, or comically float away into space.

With time now a precious and limited commodity, the audio also suffered a tragic cut. The recording session made in Hollywood was trashed, due to the plot having been changed so drastically. New actors were recruited, including Bill Johnson who returned to the role of the Guardian, replacing the actor who had co-opted the villain’s role in the first recording.

The work became frantically hurried. The writers had to write new lines of dialogue and give them to the voice actors on the very same day for immediate recording. Once a line of dialogue had been captured, there was no time to change it, even if an error was detected.

The expensive animated sequences created during Ultima IX’s second revision were reused in part, but in different contexts. Wherever possible, they were disassembled and rebuilt, but the operation occasionally left hints that were keenly identified by the most attentive fans. Thus when the game was finally released, the various intermediate versions of the plot were empirically made known.

There was no longer time for practically anything. Ultima IX had to go to press or be lost forever. After more than five years of development, an exodus of personnel, the transition to a proprietary library of graphics APIs (which had since begun to lose market), and four patched drafts, the team was unable to complete the project the way it should have been done but simultaneously lacked the strength to compassionately pull the plug.

Del Castillo exited Origin Systems with a prophecy told to Bill Randolph on the day of his departure, “They don’t care about this game; they’re just going to shove it out the door, unfinished.” His prophecy would soon be fulfilled.

^[1] Garriott, Richard (2017). Explore/Create (Chapter 16)

^[2] Ray, Sheri Gardner (2013). “Ultima VIII: The Lost Vale/Arthurian Legends RPG Interview” by BuckGB

^[3] McShaffry, Mike. “I Never gave up on Ultima IX.” The Secret of Game Coding: Game Coding Complete Guide

^[4] Randolph, Bill. “An inertia of legend,” interview with Ultimacodex.com.

^[5] The namesake of the Canadian singer-songwriter and guitarist.

^[6] Del Castillo, Ed (May 6^th, 1998), reply to a fan.