Roguelike column thumbnail ['@ Play' is a monthly column by John Harris which discusses the history, present and future of the Roguelike dungeon exploring genre. Following a profile of the devnull Nethack competition, here's an interview with the competition organizers.]

November 1 marks the beginning of the tenth-annual devnull Nethack tournament. A couple of days ago we provided an overview and took a look at the trophy structure. I asked a few questions of the co-creator and primary maintainer of the tournament. Here are his answers. Thanks to the Bandy brothers for taking time away from the tournament preparations to give us the story!

Tell us a bit about yourself.

My name is Robin Bandy, I'm 37 years old and married; my wife has played NetHack, but is not exactly excited about how much of my time the Tournament takes each year though she's been extremely patient about it. I grew up on a ranch in southwest Colorado, but I've lived in the hills in east Oakland, California, since '97.

Though my college degrees are in anthropology and history, I've been a professional geek since '94 and have been freelance since '97; most of my work these days is as a consultant running the server farm that runs, and for Ziff Davis Media.

My other main interest after my wife and my geekery is making hard apple cider (and a variety of other fruit wines as well as a small amount of beer); we're fortunate to live in a part of Oakland that was an orchard 100 years ago, so most of our neighborhood has semi-wild apple trees which are great for ciders. Cider bottling time is usually about halfway through the Tournament, which tends to liven things up a bit. ;-)

His brother, who runs another tournament server and has been involved with the running since the beginning, adds:

My name is Matt Bandy. I am Robin's older brother. I have a PhD in Anthropology from UC Berkeley and am an archaeologist who works in Peru, Bolivia, and the United States. I live in Boulder, Colorado. I had the idea for the tournament many years ago and co-wrote the code (with Robin) for the first few years. Robin has since taken the over maintanance and development of the tournament and it really is his baby and it has been for at least five years now.


How did the tournament get started?

That was actually my brother Matt's idea; as I remember it, it was basically a case of him saying "Hey, you know what would be fun to do?" and it just kinda snowballing from there.

He and I co-wrote the first version of the control server, and he wrote most of the original game server kit on his own; we still use most of his game server code, but we lost the original control server several years ago when that server lost its primary and backup disks in the same night so I had to re-write all of it from scratch. He runs a game server each year now, but with two kids that's about all he has the time to contribute.

When did the tournament first become really popular? Was it around the time Slashdot first linked to it?

The first time Slashdot posted a front-page link (in 2001) almost doubled the number of players, it's true, but the biggest year was actually the year after that. We don't have anything except the scoreboard files from the years before 2002, but I think we had ~400 competitors in 2000 and 800 in 2001; in 2002 we had 1071 players and haven't had a year that big since.

The time we consider the Tournament to have really become a staple in the NetHack community was with the release of 3.4.3 when the release notes file opened with the sentence "Now that the November tournament period is over, it is time for the NetHack DevTeam to make NetHack 3.4.3 available."; according to VersionTracker that was on 2004-01-08.

This year, actually, we're hoping to be the largest; I'm even trying to write a press release. ;-)

How well does the distributed game hosting software work? Were there any interesting technical challenges in getting the servers up and synced?

It actually does a pretty good job, though the rsync commands could stand a bit of debugging; the cron jobs throw errors when the source directories are empty for example.

The niftiest thing in it, in my opinion, was the way Matt set it up to work via ssh remote command execution so that the one control server can execute scripts on the game servers to get information from them; combined with a clever bit of Perl that lets us use flavor-specific Perl libraries for various commands means that the same script can be called on the various game servers and work the same way even though they're running many flavors of Unix.

I've gone on to use that style of remote command many times since; it's a staple for the entire admin structure on the servers that I built and run for, for example.


The trophy system came about as a fairer way to judge playing skill than the old score-based system, which is vulnerable to certain score-optimal playing styles that aren't generally considered normal play. Would you like to discuss the thinking behind the current trophy setup?

The grand prize trophy is basically a stunt (a streak of 13 winning games covering all races, roles, alignments and genders), and now that more players almost managed it in 2007 we are considering how to replace it with an even more outrageous stunt; that's the kind of thing that should be there at the top.

The major trophies, other than First Ascension, are all intended to recognize different play styles that have developed in the community over the years; the Recognition trophies are also aimed at that, but we extended them down to include folks who (like myself) cannot reliably ascend.

OK, another one of them is also not really based on a play style but rather on an aspect of the game: it kills a lot of characters, so the Most Unique Deaths trophy came in as a joke originally; now, though, it has actually developed into a play style that (as you mention below) abandons the concept of "winning" completely.

The minor trophies continue to reflect NetHack's score, since it would be inappropriate to just ignore it. Many folks have made very good cases that it doesn't really reflect the degrees of skill between the top players, though, which is why we brought in the major/minor distinction in the first place.

The Clan trophy is a bit of a joke, as the whole clan system is (Hi EIT!). It would be a lot more meaningful except for the fact that RGRN overshadows everyone else by such a huge margin. This is definitely an area that could use some serious re-imagining.


One of the most interesting trophies, although one that a player has to almost abandon going for wins to shoot for, is Most Unique Deaths. For 2007's Most Deaths trophy, player theta achieved 74% of all the deaths categorized in the game, for a little more (according to my math) than 80 death types. Another neat trophy is the "Death by Trickery" consolation prize. Are there any other clever new trophy types (or any new trophies at all) coming this year?

I really like the Most Unique Deaths trophy, and I really enjoy watching it when it gets competitive; some of the deaths that have been posted can only be achieved by getting a game up to the point where it could be ascended and then doing something stupid like eating the corpse of Famine.

Trickery is actually the game's internal name for what happens to you when something goes wrong with the game files that makes the game give up completely; this is usually caused by an admin screwing something up, so we adopted it as the apology for when something we do kills off a character.

We do have one more major trophy in mind and I've actually done the artwork for its icon, though it won't premier this year; it will be named something prosaic like "Most Extinctions", but I think of the icon image as being "ASCII Chief" since it's based on the iconic helmet from Halo.

(Note: this will probably be a trophy for the most complete "extinctionist" game. Nethack keeps a count of the number of monsters of each species that are generated, and when the count hits 120 most of them will no longer be randomly generated except in special circumstances. Extinctionist players seek to do this to all monster types that respect extinction, which tends to make for very long games.)

For several years now marvin, a.k.a. Christian Bressler, has won the grand prize, Best of 13. Last year sawtooth game him a bit of a run for his money, tying for most ascensions overall at 14. Now that multiple players are getting to the point where they are competitive for the grand prize, is it possible that you'll have to come up with some other way of measuring playing skill?

Yep; we're really pleased to see more players rising up to those levels. The closest anyone but marvin has gotten to the Best of 13 is sawtooth's 11 from two years ago, so there's probably some life in that stunt yet.

We don't have any ideas in mind for what would replace it, but we've gotten several suggestions. If (or when) we do bring in a new grand prize we'll keep the Best of 13 active, probably by downgrading it to being a major trophy.


Would you mind telling us a bit (officially) about the Challenge system and player reactions to past challenges? How do players "opt out" of a challenge? Are there and difficulties with the challenge patches accidentally changing the logic of the vanilla game, even if the player opts out? Would you like to drop any hints about this year's challenge?

The Challenge trophies I added in 2005 when I had to write a new control server from scratch; they (like the Recognition trophies) were intended to open the scoreboard up to some new folks who (like myself) wouldn't be able to compete at the levels the old scoreboard required. By adding this additional dimension to the Tournament, and by making it one that the serious ascenders would probably avoid, it gives a lot more people a chance to get onto the scoreboard and a lot of new ways to play.

This year's Challenge really expands on that, to the point that it has its own separate scoreboard in addition to the list of players who've completed it that past Challenges have had.

They were also intended to be a bit of an apology to the players, since the old system was wiped out about a month before the Tournament was due to start and I wasn't at all certain that I'd be able to get the new system built up enough to start on time.

Some players have been upset that this changes the Tournament from being a vanilla NetHack build, but from the beginning we've made them optional and I've gotten a lot better about actually limiting their impact to players who accept them; I think I tracked down the last bug in the Grue code this summer that was exposing it to non-Challenged players.

The Grue and PacMan Challenges have been the most complicated, since they both involved creating new levels and changing how light/dark worked for the Grues and overriding the vision system completely for the PacMan level. In both cases many characters (some of whom weren't on the Challenge) died in the first few days from bugs in my Challenge code and players have been rightly annoyed about that.

The basic idea of each Challenge is a narrative, hung onto a single feature of NetHack, which will be unavailable until the player completes the Challenge; last year's PacMan Challenge is a good example: when a player tries to eat a fruit, they'll be offered a Challenge (with nothing to say what it is; the Challenge must be entered into blindly) and the choice of accepting the Challenge, declining it or ignoring it for the rest of the Tournament.

Players who ignore it will not be offered it again in that year, players who decline it will not be offered it again in that game, players who accept it are given a little story about "the necromancers Blinky, Pinky, Inky and Clyde (the four greatest villains of their age)" since I'm very fond of stupid jokes and I also like to make sure there are enough clues around for players to figure out what the Challenge is going to be.

Once the player has accepted, they will be prevented from eating any fruit in the Game until they either complete the Challenge (by playing a NetHack-ified PacMan game) or log onto the web site to block it from there.

Hints? OK, two words: "Bizarro Orgasmatron" ... ;-)

Any plans to expand the tournament to cover notable variants, like SLASH'EM, SporkHack or Nethack Brass? How about other popular roguelike games like Angband or Crawl?

I would definitely be up for that, but I'd really need to do it by working with someone from each game's community who could be responsible for that game's presence; in addition to having to patch NetHack to get it to log the data we need for the trophies, I really think an important part of the Tournament's acceptance has been that we try to build it in NetHack's style as well as we can and I don't really have the experience with any of the other major roguelikes to do that for them.


How about leaving a devnull Nethack server up year 'round?

It's said on the web site for many years that we would do that, and we really should either do it or delete that line. ;-)

At this point, though, I'm leaning towards not. Unless NAO is short on capacity, they do an excellent job of providing a public playground and I really can't take the time to either adapt our system to running a regular server or learning how their system works.

If they do need additional capacity, or if they disappear for some reason and no one else can step in, we would certainly dedicate at least one server to being a full-time NetHack service but as long as we're not needed for that we'll keep our resources focused on the Tournament.

I am, however, working on another roguelike project that will involve providing a server; I mentioned on a few weeks ago that after 10 years of running this Tournament I've finally decided to try my hand at developing a new roguelike on my own and (though that won't be a /dev/null project) I will definitely have a public server up for it.

Would you want to say something more to @Play's audience?

It's not a game unless you can lose; in a great game, losing can be at least as much fun as winning.

In my opinion that's the core of what the roguelike games contribute, and it's a lesson that could have made a lot of the mediocre games out there genuinely great.