Chapter 8
While Harrison kept elegance and organization at the top of his priority list, he made plenty of other changes to the code. He added inscriptions on items that made their abilities clearer, an options menu where players could configure the game more to their liking, a bot—known as the borg—able to play the game of its own accord, and, with the help of a friend and fellow programmer, tighter line-of-sight code engineered to benefit players stuck using older hardware.
"Run through the town and watch the floor light up, and you are seeing the most obvious use of the new code. But it provided a massive efficiency boost which was particularly important on lower-end machines. Programmers these days are spoiled by fast machines."
Establishing a team of maintainers allowed Sidwell to cover up what many might have viewed as a weakness. "The big secret is that I'm a crap player and I'm bad at balancing," Sidwell admitted. "My main attempt at re-balancing the game—around the 3.1 series [of updates]—was, if not a failure, definitely an indication to me that I should try to find other people who had a better sense."
With a team more interested in addressing game balance, Sidwell could stick to streamlining the game's code by building on top of the spring cleaning Harrison had performed, and adding in more variation to present an even more unique experience to players every time they ventured inside Morgoth's fortress.
For a long time, the majority of players were hanging around levels where potions that increased stats appeared. "When we find tactics we don't love, we'll subtly incentivize it or remove incentives from doing it. We don't want to outright ban it or say it's cheap," Osheim clarified. The devteam countered item-grinding tactics by widening the curve on stat-gain potions, making them appear more frequently at lower depths. In recent years, empirical data has confirmed a predilection for deep-diving, moving to the bottom of the dungeon as quickly as possible in order to find better treasure. Most players understand that the trick to survival when deep-diving is to avoid monsters whenever possible.
Transitioning to a team-based approach didn't go off without a hitch. "I happened to join around the same time as a couple of other guys, and the game entered a period of very, very rapid change, which had its pros and cons in the community," Carr recalled.
Before the advent of the World Wide Web, Usenet, a discussion board where users could post and reply to notes, was one of the primary methods of communication online. As the Web spun farther and farther, Internet connectivity became cheaper. Communities picked up stakes from Usenet and relocated to websites that made it easier to host files. Over 2009 and 2010, Angband's players went with them.
The move to Github and the Web resulted in an explosion of productivity for the devteam. New versions appeared for download almost every night. Realizing that they needed to delineate the vanilla version of the game from experiments that may or may not be integrated into the codebase, the devteam created separate branches. There was the main branch, which hosted the latest version that anyone could download, play, and modify. Then there was a development branch, where devteam members could conduct experiments like mad scientists. It did not matter if their tinkering broke vital components, since their experiments took place on a separate branch than the main codebase.
Still advocating transparency, Sidwell made the development branch visible to the community. Community members were welcome to download the works-in-progress and comment on the embryonic ideas modeled therein. The forums hosted roughly three hundred active users at any given time, and only a fraction of those bothered to climb up to the development branch and sample its fruit. Those who did had a lot to say.
"The people on the forum thought that what was in the development version was the way the game was going," Carr admitted. "There wasn't a lot of clarity of, 'These are development builds, so stuff can be reversed and re-balanced at any time.' There was a lot of arguing and discontentment about some of the changes. Some of us, including me, spent far too much time and energy arguing on the forum instead of developing the game."
One of the ongoing experiments on the development branch centered on rebalancing ego items, special weapons and armor that sported special properties. The devteam was trying to determine the rate at which ego items could be generated. At any given time in the trial-and-error code, ego items were as bountiful as tiles of stone on the floor, or as rare as a dragon on level 1. Believing that the code foreshadowed things to come in updates, members of the community ranted and raved about the rate of items drops. The devteam's transparency had backfired.
Eventually, the issue was sewn up. The devteam created a third branch called Angband 4, overtly demarcated from the 3.x series of updates. Community members grew to understand that only the code on the 3.x branch would unquestionably be found in official releases, while Angband 4 was more like a secret lab where the Dr. Frankensteins of the devteam created and killed features on a whim.
Gradually, outcry on the forums died down. Rather than members of the devteam needing to shout over everyone, community members who had been around to see the experiments explode in everyone's faces were able to explain how the branches worked to newcomers. For anyone still grumbling over changes in any branch, there was always an easy fix available: download the code and create a variant.