If you haven’t read part one yet, start here. It will get you up to speed.

When Middle-Earth was in Oakhurst its producer and the general manager of the studio were both big supporters of the project. They were on board with how much work it would take and what the game would be like when it came out. Unfortunately neither of them relocated to Bellevue so after the move we didn’t have either of these champions to make sure the game was well received in its new home. This was the first, and arguably the biggest, of our problems once we made it to Washington.

The general manager part was played by the manager in charge of Sierra Studios. Both Middle-Earth and Babylon 5 were put under him. Let’s call him Sam, even though it is not his real name. He immediately started looking for a producer for our project, and a couple months later hired a guy that we can call Boromir. Now Boromir had two big problems: he didn’t like our game, and he was a lousy manager. Boromir spent all his time on the project traveling to conduct focus sessions at conventions so that he could use the resulting data as a tool to convince Sam that the Middle-Earth Online that was underway was not worth making.

My visibility into these events was somewhat limited. I was a worker-bee programmer on the team, and not invited to any Big Important Meetings. What I can tell you is that I never saw any scheduling effort from Boromir (and we desperately needed a schedule.) When I tried to get us access to the bug tracking tool from QA, he wasn’t interested enough to rattle any cages. The time he was in the building, he never came out of his office. I doubt he knew the names of any of the people below the lead level on his team. As far as I could tell all he did was disagree with everything that was coming out of the design team.

Not that everything that came out of design was golden. Some of the ideas actually survived into the game that Turbine eventually shipped, but some were what I like to call “crazy.” The biggest of those was perma-death. The high level design here was that any player would be able to work up the nerve to commit murder by way of lesser crimes. Eventually they would be able to permanently kill another player’s character. Certain high level monsters would also have the ability to perma-kill a player character. To be fair, this was 1999. Everquest hadn’t launched when we relocated, and things like perma-death were considered debatable. In retrospect, though, that one just seems crazy.

The other two big points of design contention were player psychology and player capture. Psychology was the solution to newbie ganking. Whenever you were threatened with a fight that was far enough above your level, your character would automatically run away. Player capture was the notion that certain creatures would capture your character in such a way that they would need to be rescued by other players. Neither of these was nearly as crazy as the perma-death, but taking control of a player’s own character away from them is always a risky thing.

After management infighting, the other big problem we had was actually that we had too little management. Our lead programmer, who we’ll call Frodo, was also our lead designer. Each of those is more than a full time job on a game the size of Middle-Earth. Trying to put both duties into a single person is just silly. He didn’t have the time to devote to either so both halves of his job suffered because of it. This one guy had 7 programmers and 2 designers reporting to him. About 4 people of any sort is generally considered to be enough to keep a manager from getting any non-management work done. And at the same time he was interviewing producers and then, after Boromir started, fighting constantly with him over fundamental game design.

About six months after our relocation to Bellevue it was clear that things were not going well. Sam called a meeting where he told us that there would be another meeting one week later to announce our fate. Apparently the Babylon 5 team had a similar meeting. We spent our week playing Re-volt and Rogue Spear. Nobody was interested in working on a game that was about to go away. One week of later he announced that both projects were cancelled and that we were all being laid off. To the public they said the development was being “restarted”, but the truth was that Boromir was the only person left on the Middle-Earth team, and no one was left on the Babylon 5 team.

It sucked to have the project cancelled, but there was a big silver lining. While we were deciding whether or not to relocate to Bellevue one of the artists on Babylon 5 asked, “So what happens if we relocate and then you just lay us all off anyway?” He was assured that wouldn’t happen, and to back up that guarantee they offered the existing severance package plus an additional three months of severance if we were laid off within a year of the relocation. In the 9 months I worked for Sierra I made about two years worth of salary between all the bonuses and the huge severance package. Sam did not look happy when he had to tell us they would be honoring the deal and giving us all that extra severance, but after all the crap Sierra put us through, I think we deserved it.

(You might think this is the end of the story.  Well not quite.  You can find the end here.)

All this talk about Sigil reminds me of the bad old days when I worked at Sierra on Middle-Earth Online. Much of the crap that the people at Sigil are slogging through right now is the same crap that we had to deal with back in 1999.

I started at Sierra in December of 1998. Though I had been out of college for 5 years, this was my first game industry job. I had been applying to game companies regularly for a couple years at that point without much success. In fact, the Sierra interview was my first with a game company. The position was to work as a programmer on a new online game called “Middle-Earth Online”.

The Middle-Earth team was running at Yosemite Entertainment in Oakhurst, California. Oakhurst is a little mountain town half an hour from the south gate of Yosemite. It is actually where Sierra started way back when, and though corporate headquarters moved to Washington State in the mid-1990s, many of the developers stayed behind to work on games. This was the studio that produced nearly every game that Sierra is famous for, including King’s Quest, Quest for Glory, Space Quest, Police Quest, and Leisure Suite Larry. All the early 80s stuff you read about in Smart Bomb and Hackers happened up in Oakhurst too. Some time after HQ moved away, the original Sierra On-Line studio renamed itself Yosemite Entertainment.

I was a huge MUDder back in college, and was looking to networking as my ticket into the industry, so this was nearly a perfect fit. There was no way that I could pass this offer up, even though it meant less money and moving to the middle of nowhere. Both of these things were actually a bigger deal for my girlfriend-at-the-time than for me, and it didn’t help that I didn’t tell her I was looking at jobs outside of the bay area. But after a few huge fights we decided to move to Oakhurst so I could take the job. (In retrospect, I probably should have avoided the next two years of painful breakup and moved to Oakhurst alone, but that’s another story entirely.)

From the start I could tell this was the job for me. In the first few months I worked on getting the MSQL database up and running, experimenting with terrain processing from high level maps, and reading all the Lord of the Rings books. Every day lunch was filled with chatter about games, and every evening was a LAN party on the machines in the office. Elsewhere in the same building were teams working on a Babylon 5 space combat game, and a squad tactics shooter called Navy S.E.A.L.S.

It seemed like everything was going well until… One Monday morning at the end of February I arrived at the office, as usual, a bit before 10am. The first clue I had that something was wrong was that nobody was working. Everyone was standing around talking. It seems that all the Maya and Max dongles were missing, so none of the artists could work. Someone had come in over the weekend and taken them all. The internet connection was also down, along with the email server. No one from IT could be found, which wasn’t entirely unheard of since IT support at Yosemite Entertainment was typically pretty bad, but not being able to find any of them when all the important servers where down was strange.

I had only been in the game industry for about three months at this point. I wasn’t familiar with the constant churn of companies that I now take for granted. Many of the other people on the team were freaking out though. The artists were copying whatever files they could get their hands on to one of the few machines with a CD burner and burning CDs. They wanted to make sure they had portfolio material ready. The programmers followed suit with the source code, though it’s not clear exactly what they were going to do with it. It’s not like anybody asks a programmer for a portfolio.

There was an all hands meeting scheduled for 10am that morning. It had been announced on Friday and was advertised to be about “February birthdays and Havas update”. Now speculation was running wild about what “Havas update” could mean. Was Sierra’s new parent company going to shut it down? Were we all out of a job?

Ten O’clock rolled around and we all filed into the big conference room. They laid out typical morning meeting junk food (doughnuts, muffins, etc.) at the back of the room, and I think there may have even been a cake. Nobody was hungry, and we all ignored the food. Pretty soon a guy nobody knew stood up and started talking.

His name was Bill something or other, I think. He was down from corporate, but not directly in the chain of command. He pretty much cut to the chase and said that they were closing the studio and that some of the people in the meeting would now go upstairs for a separate meeting. At that point he started reading a list of names, and it was obvious about five names in that this was the list of keepers. He read the names of almost the entire Babylon 5 team and about 2/3 of the Middle-Earth team. We got to walk past all the people who were about to be laid off and go pack into the second largest conference room to be offered relocation to Bellevue, Washington.

Everyone was pretty upset. Some of these people had been with the company twenty years, and most of them had been there ten. The Middle-Earth people who didn’t make the cut were mostly artists from Sierra’s 2D adventure game days, plus one programmer who seemed to have been laid off accidentally. It was clear that this was seen as an opportunity to shed some employees who weren’t as useful in a new 3D era.

The people who were laid off were given two months of “notice” period plus an amount of severance that scaled with both their salary and the length of their employment. The people who were offered relocation were given a choice between either that severance package or a $10,000 signing bonus, a 10% bump in pay, and 100% paid relocation. I don’t know what anyone else was offered, but my three whole months of employment plus my $60k salary would have resulted in something like 5 weeks of severance.

At the time, they didn’t tell us why Yosemite Entertainment was closing, but everyone had a theory. Quest for Glory V had shipped earlier that year and bombed. It cost way more money than any other Sierra game ever had, and the Yosemite studio head put a big political bet on its success. Also, Sierra (or at least its parent Cendant Software) had recently been purchased by Vivendi/Havas and renamed Havas Interactive. Perhaps this was an opportunity to cut costs and “save money” through some funny accounting tricks that were only available at the time of the merger.

Whatever the reason, we weren’t alone. Dynamix was closed the same day. Berkeley Systems (makers of the very popular You Don’t Know Jack series of party games) was another Havas company and was closed not long after that. Bill something or other was laid off two weeks later when Sierra corporate decided they didn’t need that much middle management now that there was nobody left to manage. Sierra was in the process of imploding and we were just part of that process.

So that’s the story of the first three months of my time on Middle-Earth. I’ll leave the exciting tales of horrible mismanagement for my next post. (You can find part two here.)

(Seems like MyOGDC should be some sort of open source middleware, doesn’t it?)

OGDC was a very good conference. This was the first year, so it was pretty small, but that just meant more contact with the people who were there. Even the sessions that seemed like they might suck were quite good. I’m thrilled that I was able to attend, let alone speak.

Here is the high level summary of the stuff I saw at OGDC.

---

9am Thursday - Writer: the Red Headed Stepchild of the Videogame World

Our own Jess Lebow (Or as he’s known among the Dev-butt-kicking beta community, Captain Elbow) talked about what it’s like being a writer in a game company. His talk went well, and later I talked to a writer from another company who said a lot of Jess’s points really rung true for him also. He outlined a bunch of ways that a writer can contribute in a game company outside of writing the in-game text. Those include:

1. Acting as an editor for PR text like press releases, ads, and major news posts.
2. Acting as an envoy between design, art, and programming and to give these people an idea of the context of the art, system, or code desired.
3. Naming things (though Jess didn’t really seem to like that as one of the things a writer can do.
4. Developing the “internal story”, which is to say the background that is essential to telling a coherent story but is never actually shared with the outside the world.
5. Additional fiction for the web site, exclusives for fan sites, etc.

The talk wasn’t really aimed at me (and I actually heard it a few weeks ago in the office) but it seemed like it would be pretty useful to a fellow writer.

---

10am Thursday - Working in the Data Mines: Uncovering Gameplay Gold

Darius Kazemi gave a talk on metric collection and interpretation in MMO games. I had dinner with Darius on Wednesday and talked to him on this subject throughout the conference, so I’m not completely sure all of this came from the talk itself.

He was at Turbine during the development of D&D Online and LotRO analyzing metrics on both games. They did a fair amount of collection on DDO the “wrong way” so for LotRO they started with the database schema they wanted to query from and then figured out how they could log into that database from the code. He left Turbine a couple of months ago to start Orbus Gameworks, which is a metric collection middleware startup.

Darius advocates storing all sorts of game data in easily-queried tables even if that isn’t the native format of your table. His “character” table has one row per character with all the stats for the character as columns. This can be combined with a second “character history” table that contains the changes made to the character at each event. This lets you re-construct the character stats at any point in the past with a fairly simple query. Of course this diff table will get pretty big so you will probably need to keep only a short history.

He also demonstrated how “infecting” a small number of social hubs with a piece of information can be an effective way to communicate that information to your player base. A very small number of hops from the originally infected can cover the majority of your players, and the new spreads in a way that pulls the players deeper into the game.

---

11am Thursday - Automating Online Game Balance

This was a talk by one of the +7 Systems guys about how you can balance your game automatically by measuring the popularity of various game options and adjusting automatically based on those measurements. These guys sell middleware to do this, but the talk wasn’t really so much a sales pitch for their middleware as for the idea of automatic balancing in general. I came away from the talk unconvinced that automatic balancing was a good idea.When this came up in the metrics round table at AGC last year, someone suggested that it was going to nerf female night elf priests just because they were popular, even though they were popular for entire non-gameplay reasons. It seemed like kind of a problem for me, so I asked about it in the Q&A. Their answer was that it doesn’t matter why something is popular… if you want to maximize your content utilization you need to drive players to different choices through game balance. That doesn’t seem like a good idea to me.

Two things I took away from this talk are actually applicable whether you want to do your balancing automatically or not:

1. Communicate to your players that rebalancing is a part of the game. Do this continuously, and not just because you’re about to nerf a bunch of people.
2. Make many small changes instead of a few big changes.

---

Noon Thursday - Lunch

OGDC had the best food of any conference I’ve ever been to. And Thursday’s lunch was the better day. Yum!They did something really clever with lunch/keynotes. The keynotes were held in the same room as lunch, so everyone had an hour to gather into the ballroom/eat lunch, and then the keynote started. It worked really well.

---

1pm Thursday - Keynote: Concurrency

Herb Sutter from Microsoft (of The Free Lunch is Over fame) talked about the multi-core future. He estimates that we will have 32-128 core machines over the next few years. He also went into how features like futures and transactional memory are going to make concurrent programming much more feasible going forward. It was a great talk about a subject that my own knowledge is a little lacking in.

It was a little odd as a keynote though. I think it went over the heads of about 2/3 of the audience.

---

2pm Thursday - Engaging Your Community - what you can do for them and what they can do for you.

Community Managers from a bunch of games (including ours) got together to put on a panel on community management and how to get your users to help you out. These are always entertaining (which is pretty much true of any time you put a bunch of extroverts on a stage) but not terribly relevant for me.

Oh, and they always have plenty of straight talk about communities that you really shouldn’t repeat on the internet.

---

3pm Thursday - Security Issues for Third Party Games: Technical, Business, and Legal Perspectives

This was the only talk at OGDC that I regretted going to. The first half was a big long list of a bunch of security breaches that have happened in games over the past two years. The second half was a bunch of “security is important” stuff, but nothing really specific enough to act on.

I hear the Chinese Game Market talk in this slot was awesome. Wish I’d gone to that one instead.

---

4pm Thursday - LIVE on Windows Essentials

Zsolt Mathe from Microsoft talked about using LIVE on windows. There was plenty of useful information here about how LIVE works in general and on windows specifically. The big scary thing I took away from this talk was the notion that to use LIVE, even on Windows, you have to get Microsoft to approve your game. That doesn’t appeal to me at all, though I understand why they would require it since any LIVE on Windows game can earn achievements and communicate to any Xbox LIVE game. Still, the lack of a “get permission to ship” checkpoint on Windows is one of its biggest selling points.

---

5pm Thursday - It’s All in the EULA

James E. Dunstan, a lawyer, went through a EULA step by step and talked about the reasons and case law for each major section. “How boring” you might say, but it was very informative and really interesting. This guy was Mythic’s lawyer in the Black Snow case a few years ago, and has a long history of other video game and media cases. He really knew his stuff. Oh, and it was nice to finally hear a talk about legal stuff that didn’t have the phrase “but I’m not a lawyer” every 5 minutes.

---

9am Friday - Adventures in Middleware

I think my talk went pretty well. The surprising bit of feedback I got from several people was “thanks for including the prices for all this middleware.” That surprised me mostly because all of the packages I talked about publish their prices on their website. There were no secrets there, just a collection of information in various corners of the web.

I ended up in the first slot of the day, so the session wasn’t packed, but the 15-20 people who came had plenty of good questions.

---

10am Friday - Building World Class MMO While Building a Company - Cheyenne Mountain Entertainment and Stargate Worlds

Joseph Ybarra talked about how the company got started and how they are going about building a team in Pheonix Arizona, where there are no game developers to steal people from. Cheyenne Mountain is probably bigger than you expect… they are still working on their first game, Stargate Worlds, but have already spun up two additional studios to work on other MMO titles. They have tons of investment and are positioning themselves to be a big player in the MMO space. It’s really interesting to see a more traditional startup model applied to the game industry…

---

Noon Friday - Lunch again!

I ended up at a level with the +7 Systems guys and we had a nice talk about their stuff. I’m still not convinced automatic balancing is the way to go, but it was good to learn more. This kind of spontaneous conversation is something you can get at a small conference. It is much less common at GDC. It’s also why I blow off my co-workers to go sit with people I don’t work with. Nothing personal, Jess and John!

---

1pm Friday - Keynote: Games Industry 2012

Erik Bethke of GoPets outlined his vision of where the game industry is headed in the next 5 years. Here are the highlights:

• WoW will be on its third expansion, which will include both ninjas and pirates. You can get the pirate part of that way before 2012 with a certain pirate MMO, Erik!
• Software piracy will be rampant everywhere in the world, making the only viable software business models be those that are more like services than products.
• 3 billion additional people will be online with broadband. Almost all of these will be in the developing world
• Television ad revenue will plummet and the revenue from the biggest MMOs will be as big as all of television.
• RMT and item sales will be normal. The notion that the company running the game actually owns everything in the game will vanish.
• 7-11 and PayPal will have “prepaid online cash” wars with each of them providing age-restricted cash. This will allow content providers to restrict things by age without knowing who their customers are.

Some of these things sound more crazy than others. It was a good keynote overall, though. Definitely more entertaining than the GDC “our console is awesome” keynotes.

---

2pm Thursday - Culture Clash: When Security Comes Knocking

Dave Weinstein is a very dynamic speaker. So much so that it was sometimes hard to tell that there were two other lecturers.

Dave talked about the different perspectives that game developers and security professionals bring to the table. The game developers are under a huge amount of pressure to get their games done on time, and then along comes this security guy. He often arrives late in the schedule and bounces from game to game adding more work to each one. This can lead to some hostility from the game developers who have been crunching for months trying to get their game out the door. I can certainly see where that would cause a conflict. Apparently Microsoft has really started taking security seriously, and are now enforcing security work for all game projects on Xbox.

One thing I took away from this is the notion of a fuzzer. This is a piece of code that takes a mostly-legit data stream and tweaks it in a wide variety of ways to try to demonstrate bugs. That seems like something that’s easy to write but could really help track down potentially serious security issues.

---

3pm Thursday - Dissecting our Baby: AutoAssault Postmortem - the good, the bad, and all the ugly

I expected more of a game design or programming postmortem but ended up with a business level postmortem. In other words, there wasn’t as much gore as I expected.

NetDevil’s agreement with NCsoft for AutoAssault was a fairly traditional one for single player games: A bunch of milestones with pre-defined features in each one. The trouble was that this meant NetDevil had to put in some version of those features to get the milestone criteria satisfied. This kept the payments coming so they could make payroll, but didn’t necessarily result in shippable features. Each payment was an advance on royalties, which meant that no more money would come in post-launch until the game earned enough money to pay back all milestone payments out of their cut of revenue. At a certain point they just had to ship or they were never going to see a dime from the game.

Some key lessons learned/suggestions from the talk:

• Keep your publisher involved in the game on a day to day basis. Get them into your play tests, make sure they have a very clear picture of where the project is.
• Get a very small subset of the game playable ASAP and keep the game playable from that point until launch.
• Keep most of the complexity away from the player in the early game. The first part of the game should be incredibly simple
• Make distribution of builds very easy from the start. Do something like the Guild Wars in-game content downloads so you can double click on an icon to play the latest build without having to download all the content again.
• If your staff isn’t playing your game, you are in trouble. They will make all sorts of excuses (”I’m not the target market”, “I’m too busy”, etc.) but it all really means is that your game isn’t any fun.
• Don’t feel like you have to have every feature that another game has, particularly when they aren’t right for your game. (The skills in a driving/shooting game were the example here.)
• Focus test and usability test early and often. Actually take the feedback from this testing and implement them instead of cramming in all the scheduled features that you knew about before the testing.

The talk was fairly upbeat. Thanks to the LEGO project, NetDevil has survived AutoAssault, and feels like they’ve learned a ton. It was nice to see that they really don’t blame NCsoft for what happened… they took responsibility for their own screw ups.

---

And that’s all the sessions I made it to. I definitely want to go back next year. This was one of the most useful conferences I’ve been to.

As promised, here are the slides from my “Adventures in Middleware” talk at OGDC 2007. Thanks for braving the 9am time slot to come and hear me yammer for an hour. I hope it was helpful.

I had a great time and learned a lot at OGDC. I’ll have to post more on that later.

Sara discovered our secret competitive advantage. Starting a few months ago we’ve been running usability tests to try to smooth out as many rough edges in the game as possible before launch. Even though our testing is extremely low budget, we have been getting fantastic feedback on the game. The usability person we hired came on board after the first round of tests and has introduced a more scientific process to the testing. That has definately increased the value of the tests.
If you don’t have access to an expert of your own, you can still do this kind of testing. The basic approach this:

1. Invite some local gamers to the office
2. Order some pizza
3. Let them loose on the game
4. Observe

We have an FLSer with a notepad watch what they’re doing, prompt them to describe what they’re doing, and take notes on what they say and do. That’s it. No fancy cameras and no real science, but plenty of gigantic pointers at things to fix in the game. The usability person we hired has been helping out with those test, and we’re bringing her aboard to increase the pace of testing. The lab itself is just a dedicated space to do more or less the same level of testing.

I’m sure that spending tons of money on equipment and a bunch of trained professionals to use it would get us better results. Maybe those results would be twice as good, even. The cost would probably be ten times what we’re paying for our simple tests with a single trained professional. If we had all the money in the world, it would be something to look at, but for now the pizza testing is working just fine.

I wish we had started doing this years ago. There’s nothing like observing a real user to tell you where your UI sucks.