Realtime Worlds (RTW) now holds the dubious honor of being the world’s most spectacular MMOFPS failure. Baby eating was the cause.
Counting Noses
Sunday afternoon is normally a busy time for MMOs, especially in early August when school is out and summer vacation time is available for adults. APB had launched a month earlier (July 2, 2010). Its two English-language servers, Zombie and LaRocha, had a total of 1,970 players online, over 80% “paying” players (i.e., in pay-for-play regions). Six hours later, Sunday evening, the population was 2,334. Depending on play times and cycles, population in new games at busy times is 20% to 40% of total customers. This means APB probably had 5,000 to 10,000 English-speaking customers. APB supports French, German, Italian, Russian and Spanish as well. Generously speaking, the game had 10,000 to 20,000 paying customers.
This customer base translates $1.2 to $2.4 million income per year (APB charges $10/month); but only if the game maintains its users. Checking two weeks later, population on Sunday afternoon on the same two servers was 1,221. APB had lost over 1/3rd of its active players! During that time RTW’s bankruptcy was announced. Gamers don’t abandon ship purely due to news from the financial pages. In fact, within the game a frequent sentiment was “Great! Maybe someone else will start fixing this!”
Experienced game industry veterans can read these tea leaves. APB was an abject failure – a success requires at least 50k to 100k customers. A game that took $80+ million to make at a studio that burns through $20 million a year needs over 200k customers to keep going and please the investors.
A 64-bit Technical Success
By technical standards, APB is impressive. If your PC is a 64-bit multi-core monster and your 1+ Mbps broadband doesn’t drop or reroute packets like hot potatoes, the game performs impressively. You run, jump and hurdle fences in a large city district with 79 other players and hundreds of NPC civilians. Far more impressively, up to one driver and three passengers can travel together in a fast car that spins, slides, bounces and rams into other vehicles, including airborne vehicles that can launch themselves from ramps, roofs and overpasses. All three vehicle passengers can shoot, be shot at, hit moving targets while in motion.
Some people complained about the server-authoritative cars, but with a decidedly mediocre internet connection I found APB cars reasonably drivable. Like Gran Turismo or GTA vehicles, or landing a plane in a flight sim, they require a gentle touch.
A 32-bit Windows XP box can run this game, but not well. What is an impressive game on a $3,000+ Win7 PC (2.8 GHz i7 930 quad-core with 12 GB RAM and an NVidia 480 GTX) becomes sluggish, with odd pauses, on a 4-year-old $2,000+ WinXP PC (3.2 GHz Core Duo with 2 GB RAM and an NVidia 9800 GT). I suspect this is the source of many complaints directed at APB.
Given the huge success of the Grand Theft Auto (GTA) series, hard-core gamers with bleeding edge hardware “should” enthusiastically embrace such a technical tour-de-force. Unfortunately this didn’t occur. Steep technical requirements are just the tip of the iceberg. APB hit something far larger, well below the waterline.
A Design Disaster
In APB “Rating” level (R-value) is a key measure of progress. Win or lose, a character gains rating level in the game. In fact, losing a big fight with multiple players gives ten or twenty times more “rating” than winning an unopposed solo mission. The tutorial gets you to R20. After that a player gains about 2-4 rating points per hour to R100, then 1-2 pts/hr to R200, and less than 1 pt/hr to R300, etc. After six weeks only a handful of players were beyond R300. A $50 initial purchase gives you 50 hours of gameplay, enough to reach somewhere between R100 and R200 unless you use cheats and hacks for faster advancement. Grinding through enough gun battles for R-level is one of four disparate and confusing advancement tracks. It is as if the APB’s designers took the worst parts of Warhammer Online’s advancement system and made them even more confusing, something even the most cynical designer might have thought impossible.
Ultimately, all these advancement systems provide a character with personal improvements that add hit points, reduce incoming damage and increase the healing rate. Overall an R200+ character “with benefit” is about 30% more survivable than an under-R100 novice. Those same veterans also get guns with upgrades that increase damage, accuracy and rate of fire for about 30% more firepower. Combine both (1.3 * 1.3) and veterans are 1.69 times stronger than newbies. The practical effect is that in a face-to-face gun battle, veterans can’t lose unless they fall asleep. In fact, even if the newbie surprises them with a blast into their back at close range, the veteran can turn around and return fire so effectively that the newbie still dies first.
In addition, with 75 to 150 hours of playtime, R200+ veterans have experienced most missions multiple times. In the process they learned the best places to hide, the fastest routes to rooftops, the layout of multi-level shopping malls, and how to use this terrain to best advantage. There is little chance a newbie can find any position of advantage against a veteran. There is every chance the veteran will find a superior position and take the first shot.
Next factor in the various hacks developed by enterprising entrepreneurs during beta. These went on sale (through the black market) the day the game launched. Aimbot hacks let players automatically zero-in instantly on a target. Wallhacks allow players to see and shoot through walls. Most recently gunhacks give every weapon the longest possible range. Realtime Worlds never developed good software solutions for these. Instead, they relied on players making video recordings of hackers, which Realtime Worlds CSRs (customer service representatives) would presumably view and judge who as cheating. Needless to say, this is the slowest and most costly way to fight game-wrecking hackers.
All these problems pale against a truly vast design flaw. APB allows players of ALL levels and hours of playtime into the same instance. Guess what? Players at R200+, with all their advantages, constantly slaughter low rating players. Those lower players don’t just lose a match or two. They lose miserably, outmaneuvered and outgunned, hour after hour, day after day. At 2-4 hours per day it would take over a month to reach R200+ and be “almost” competitive.
The typical gameplay arc PC and console games, “try, lose, learn, win,” takes ten minutes to an hour or two. The gameplay arc of APB is “lose every match miserably for a month, and then win occasionally.”
I witnessed the effect of this is an APB guild during the first month. They began as an enthusiastic group of role-players. Then they lost miserably against seemingly “unkillable” enemies who regularly outmaneuvered them and insulted them. They moaned about hackers and developed a “celebrate each kill” mentality. They tried to ignore mission defeats. But it’s hard to lose most of the time. Within the month I saw groups tending to hang out, chat and goof around. They spent less and less time taking missions and shooting weapons. More and more members drifted away. Last time I looked participation at my regular gaming times had dropped to half, with one or two more vowing to quit that week.
Baby Eating
A game design is a “baby eater” if high-level players constantly defeat low-level players. APB is a classic example of this. Incoming “baby” players experience nothing but defeat as veterans tear them apart. Despite claiming that a special “threat” system would create “fair” matches, the actual system completely failed. Vastly unequal matches were commonplace in APB. This resulted in no positive word-of-mouth encouraging games to try APB. Instead, discouraged novices spread “bad vibes,” in the form of complaints about everything from real culprits (such as the matchmaking system) to irrelevant issues (a lack of “realistic” gun recoil).
In the final weeks before bankruptcy, Realtime Worlds desperately patched and “fixed” APB. Unfortunately these were minor tweaks to weaponry, outfitting and matchmaking adjustments that did nothing to prevent “baby eating.” Perhaps the senior RTW designers were so in love with the original concept that they couldn’t see the horrible reality. Perhaps there wasn’t the time and resources to make wholesale post-launch changes that fast. The inability of those designers to see the problem during beta was fatal. After launch they were reduced to rearranging deck chairs and conducting one last tune from the sinking fantail of the Titanic.
Well, Mr. Smarty Pants, What Would You Do?
About 18 to 30 months ago, the proper decision would have been launching a preliminary version of the game with core gameplay. A small, simple game would have been appropriate: allow characters with guns fight on foot in one reduced-size city district. The subscription business model with very modest fees ($3-4/month, instead of $10 or $15) would encourage realistic user behavior. Such a game would quickly demonstrate whether players were coming back or more, or leaving in droves. If they were leaving, developers could experiment with everything from gameplay to matchmaking to business model until something worked. Until gameplay builds a proven audience, no amount of AAA “chrome,” from vehicular travel to character and vehicle customization, to UI polish, will make an MMO loser into a winner.
Unfortunately, outside the field of social network (“Facebook”) games, few MMO developers and publishers have the courage to perform real-world billable betas with “incremental” development. Management usually lives in the fading mindset of “big launch” boxed products. I’ve encountered innumerable game marketing managers and VPs claiming that sufficient budget can make any MMO a success. Similarly, I’ve met countless game designers convinced that his or her grandiose design vision will be a smash hit, if only they are given the time and resources to “do it right.”
Even if game studios and publishers avoid these traps, outside financial backers and venture capitalists may insist on it. The “money men” are not game industry experts. They rely on the advice of others. This source of this advice ranges from the teenage gamer next door to “industry experts” who “graduated” years ago from studios and publishing companies into the ranks of paid consultants. Sadly, too many of these experts remain in a time warp, believing that modern online games must be sold like “big launch” PC and console games of the 1990s.
Another pernicious influence on game development is selling investors using a “cult of personality” gambit. In this enthusiastic pitch-men (or women) puff the reputation and trot out well-spoken, well-known industry figures to give the company or game sufficient “gravitas” to land another $10 or $20 million in investment. Examples of this include John Romero at Ion Storm (Daikatana), Will Wright at EA/Maxis (Sims Online), Brad McQuaid at Sigil (Vanguard), Richard Garriott at NCsoft USA (Tabula Rasa) or Dave Jones at Realtime Worlds (APB). In reality, as some of these people quickly point out, it takes a dedicated, skilled and experienced team of 50+ to make a great MMO. A “front man” who spends most of his time with investors has little opportunity for more than a token influence.
The best road to industry success is to start with an honestly led, well-run, experienced development team. Release early and iterate toward success on a five year or more arc. If your business model relies on recurring revenue, like most MMOs, bet on gamer word-of-mouth and a slow marketing drumbeat over “big launch” events.
If faced with rescuing APB today, a different strategy is needed. My first move would be splitting development into two teams: ‘live’ and ‘relaunch.’ The live team would be tiny, concentrating on dealing with hackers and making tweaks to “maintain the faith” of those still playing. The great majority would go to the relaunch team. Their prime goal: find and execute the fastest possible solution to “baby eating” gameplay.
One possible path is an instance system that absolutely prevents higher-powered players from fighting lower-powered players. How? Segregate players into low, middle and high level instances. Absolutely prohibit higher level players from entering lower instances. Lower level players can “play up” into higher levels whenever they wish. As lower players advance, they are gradually “pushed upward” into higher level instances. Of course, this requires a good way to measure player ability, which never easy in a team-based FPS.
Another, easier path is overhauling the advancement and equipment system. Give novice players the absolute best single weapon at the start, as well as top-level character enhancements for the best survivability. As a player advances, give them access to alternate enhancements and alternate weapon options. Allow high-level players to explore different options. This is not a new idea. It was used with success by Planetside, one of the first MMOFPS games.
In either case, the new “baby savior” system would need a new server to test and ultimately house the new gameplay. A solid relaunch effort would require new “content,” such as a new region of the city, new character clothing options, new vehicles, and new weapons, to help interest previous players and hold onto existing players. Combating hackers and “baby savior” gameplay will not make APB an overnight success. The right fixes simply allow a profitable game to emerge from the current wreckage. Over time wise stewardship and a careful attention to gameplay and features could breed success, just as CCP grew EVE Online from 50k to 500k subscribers in seven years.