Short-Term Luck and Long-Term Skill

Having finally found my way back onto my own website (a surprisingly challenging task these past few days), I present a blog entry in two parts. The first shorter part will discuss the current status of URR (short version: development resuming in under a month!) and the second will be a more theoretical bit of rambling about games, luck and skill, “investing” time/effort into a game, and similar (short version: roguelikes are sort of like poker. In a way. Kind of).

I’ve been thinking about what I want the next release to entail. Roughly speaking, I had several major objectives for this release: ziggurats, ziggurat puzzles & artwork, history and civilization generation, and the encyclopedia. I’ve now decided that I want to prioritize the gameplay aspects of the first two of that list over everything else for this release. I would estimate one/two months ZG2work remaining in ziggurats, and development will be resuming towards the end of May. Despite the progress made on flag, religious icon, creation myth and history generation, I’ve decided to leave these for 0.4. I want to get some gameplay out. I want people to see what I’ve been working on in URR, get some feedback, and actually try to release two versions a year. Lastly, I realize that the flag/ religion/ history/ myth gen aspect is (like everything…) larger than anticipated, and really merits a release in itself. So that’s what it’s going to get. This means 0.3 will have a lot of small changes/improvements, but will be primarily focused on ziggurats, exploring them, solving the puzzles within them (which get harder with each floor), and last – but not least – finding a particular kind of item atop the ziggurats that will tell you a little bit about the future of the game. I’m not going to say anything more about that – you’ll have to master a ziggurat to find out.

Now onto this post’s scheduled dose of video game pondering. I’ve long thought there was a strong corollary between roguelikes and the various games of poker, so I’m going to discuss a few of the strong simiZG2larities I think they have and how they relate to roguelike game design. Some time in the past now, I played poker semi-professionally in order to supplement my income for about two years. In that time I believe I logged something around a million and a half hands online, though I do no longer have the datamining files any more so it could probably be +/- 200k or so. Regardless, there are three central ways I’ve always thought roguelikes scratched the same itch as poker, and I’d be keen to hear what people think about these comparisons. The first is in relation to luck and skill, the second is in relation to “investment”, be it time, money, effort, or whatever else, and the third is in relation to the skills they demand.

Poker has a very high short term luck aspect. Every single hand dealt is completely random. What this means is that if you put the world’s best player in a room with the world’s worst and have them play a single hand, the world’s best would win more than 50% of the time, but it would be far from 100% because so much of that single hand depends on that particular deal. This is why poker players, particularly online ones, don’t count results from single hands, or even generally from single sessions, but across months, or hundreds of thousands of hands, etc. Over that kind of time span, the number of strong and weak hands anyone is dealt, the number of times a player is in a particular position, the number of times the player is at a table with strong players or weak players, and every other factor that affects a given hand will even themselves out pretty quickly. With such a large sample size the stronger players will pull into the lead in the long run whilst the weaker players will fall behind. Of course the weaker player will sometimes win and the stronger player sometimes lose, but in the long term the stronger player will lose less from losing hands and win more from winning hands than the weaker player. Indeed, many say this is a crucial part of the game’s appeal, and something purely mechanistic games don’t have – weaker players remain keen to play and risk their money because they will win sometimes, and this is why “professional poker players”, as a demographic, are vastly much more numerous than, say, professional chess players. There’s simply so much more money around.

I would suggest that a single hand in poker is roughly equal to, perhaps, a single floor in a classical roguelike, or a single room in the Binding of Isaac, or a single ship engagement in FTL. The elements of that floor are broadly random (though obviously within predefined parameters), and you never know when entering a single floor whether you’re going to be particularly lucky, unlucky, or neutrally lucky (were there such a term) on that floor. Equally, rooms on Isaac range from the trivial to the painfully difficult once you reach the Chest, and ship fights within a given sector of FTL have a reasonable amount of variation – though they are not labelled as such, you will encounter particularly challenging ships now and then, equivalent to unique foes in classic roguelikes or miniboss rooms in Isaac. In the short term there is a big element here, and sometimes it may overcome the player (though it reassures me that people can get 20+ game streaks in games like Crawl, thereby demonstrating the mastery a skilled player than possess), but over time the stronger player will survive that room the more often (perhaps reaching 100% in some games). What you will face in each room or floor or battle is luck, but completing a given playthrough comes down entirely to skill.


Then there’s the second comparison which I think bears mentioning – that of “investment”, both in terms of time and effort. This is less true of “ring games” in poker, but very true of a tournament. Let us say you pay $50 to enter a tournament of 2000 people. The way poker works, the top 10%, i.e. 200 people, will get paid. The 50th place finisher will get $50, the 40th only around $100, and it will quickly ramp up to a very, very high first place prize. For 2000 people to go down to 200 people could take the best part of half a day. If you enter the tournament, play, then bust out 51st, you have just spent $50, and 6 hours, for nothing. In fact, you would have been better had you not played – you could have spent the time and money elsewhere. Compare this to a roguelike and you get the same formula, at least in terms of time – you invest a lot of time in a character, and whilst upon dying you may learn something new, it’s hard not to feel you wasted your time on that character (at least for me). The reward of eventual victory in a new and challenging RL certainly offsets this, but it’s still hard to not feel some time was wasted. Unlike games with checkpoints, or the ability to quicksave, or whatever else, I would suggest that you are taking a gamble when you invest your time in a roguelike in a way very akin to a poker tournament. You risk a lot, but the reward – the satisfaction – trumps almost any game where you can save/reload. It’s up to the player to make that risk/reward decision.

Thirdly, and finally, the kind of skill each demand. In my department at the university, there are a lot of people who play chess. Though I have played a few games, won a few and lost a few, I’ve never particularly enjoyed chess. It’s too mechanistic. “Skill” in the early game and the late game, for me, comes down toZG2 nothing more than a memorization exercise of learning good openings, weak openings, how to counter openings, etc. Where’s the skill there? Whilst I admit the midgame has a lot of room for genuinely interesting, skilled play, the early game in chess is so dependent on rote learning – and a good/bad opening sets you up well/badly for the entire game – that the game as a whole just doesn’t appeal. By contrast, games with short term luck ensure this is never the case. You are constantly kept on your toes, and constantly challenged by having to deal with something new. You can’t rote learn in poker – you have to adopt a new strategy every single hand, broadly speaking. The same goes for roguelikes – obviously you have to learn the rules and mechanics, but there is minimal rote learning of what you do with them (outside “find the vibrating square” in Nethack, “collect three Runes” in Crawl, etc). Every floor and every situation is unique, and it demands at every stage the kind of mastery over the game that, in chess, is only found in the midgame. A game which makes you respond to something new at every moment seems to me infinitely more interesting.

In case it was not clear, I think these are all good things. I like that roguelikes (for the most part) avoid getting stale, or at least get far, far less stalZG2e than non-procedural games, and I like putting my time “on the line”. Naturally when I lose a Crawl character at 14 runes (sigh) it’s infuriating, but with what I now know about the extended endgame, I’m confident I’ll get a 15-rune victory at some point before too long. Equally, whilst I won’t say a roguelike “never gets boring”, it certainly does so at a point slowly than other singleplayer games. There’s also a great level of satisfaction to be gained, I think, from that kind of mastery over the procedural elements of a game, and one generally greater than mastery of mechanistic elements. That feeling that you can defeat “whatever the game throws at you”. I discovered roguelikes about the same time as becoming good at poker (the latter became “relaxation” after the former), but the two really do work on very similar psychological levels. How does all this whatnot tie into URR? Well, I am going the permadeath route, but I want death to be a less regular happening – and one the player has a lot more control over – than in, say, DF adventure mode. Short-term luck should affect your floor or current foe, but it should never turn a great run into a dead run without your control. It’s a tricky balance to get, no doubt, but there are enough games in which it works to make me confident it’s a balance I can get right.

Be Sociable, Share!

6 thoughts on “Short-Term Luck and Long-Term Skill

  1. I just wanna say that people who do 20+ streaks in Crawl are most definitely using a single or possible 2 or 3 different incredibly formulaic builds, that they have have meticulously devised through careful calculation and planning, IMO Crawl is mostly about knowing what class/race/skill/item combinations that are poorly balanced and exploiting them.

    • Very true, most roguelikes have a very pattern based early game, ToME and ADOM come to mind as well as a similar endgame.

  2. Always a good day when you decide to update! Can’t wait for .3 to come out, though I am pretty hesitant on the whole puzzle aspect of ziggurats. How exactly are you planning to make them less repetitive? Is the puzzle that unique or are you planning on multiple puzzles?

  3. I usually don’t have much to say, but I’d just like to tell you that this was a very interesting read, and you helped to make up my mind in my own design process. So thank you, and good luck with your puzzles.

  4. Sort of OT, but… Hey man, I know how tough it can be to have a crap-ton of features that are almost ready and it’s time to release something playable. Good luck, the progress you’ve made is stupendous and it inspires me to work on my own game.

  5. @ Scab – ah, but in the recent tournament, I’m sure there were people getting streaks that long with a different build each time – they weren’t allowed to cycle classes/species until every class/species had been used. Clearly they used the optimal combos for each, but there are still some species that are never the less a hell of a lot harder than others. I agree that’s certainly a factor, but it’s far from the case that anyone playing as a strong class/race combo can win (as the Crawl reddit aptly demonstrates).

    @ Grant – the puzzles are procedurally generated. There’s several hundred thousand possibilities, and that’s just for this release – I intend to add many more variations in later releases. They scale in difficulty; puzzles on low floors are pretty small, but the ones on higher floors require a lot more thought. Currently “clues” and information for the puzzles only come from one source, but as I say, I will later be adding more types of clue. This version will show off the solutions, and one form of clue (the main form), and the stuff hidden at the top of ziggurats.

    @ Ayrb – well, thanks a lot! Very kind of you : ).

    @ Eclectocrat – ha, yes, this is exactly the thing! The encyclopedia looks great, and I’d really like to show off the flag generation more and that kind of thing, but I’d rather leave that until it can be fully implemented along with civ/history gen. And cheers! I’m really looking forward to getting back to it in the next couple of weeks.

Leave a Reply

Your email address will not be published. Required fields are marked *