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.

Why make a roguelike?

Explaining why I’m actually making a roguelike seems like a logical start to this blog. So: why, having played games my whole life and having amassed a not-insignificant collection of them (including some very rare old home computer editions), have I decided to finally produce one myself? And perhaps the corollary to that is – why, being 21 and relatively computer-savvy for someone who has never studied computer science, I haven’t made one before now.

1) I’ve always wanted to.

Huh, that’s an obvious one. Surely there’s more to it than that?

From the time I first started playing games when I was very, very young, I’ve wanted to make one myself. I even tried doing this on BASIC on my Spectrum ZX when I was six or seven, and managed a truly spectacular one-minute game where you fought off alien fish, graphically based on (read: stolen from) Exolon, the game shown below.

Screenshot from Exolon on the ZX Spectrum
The graphical style I ‘borrowed’ from.

I assume there was some coherent internal narrative there, though what it was I for one can’t figure out.

Nevertheless, from then on, every so often I’d come up with a great idea for a game, or a mechanic, or a plot, or something else, and start planning out precisely how this game would function… before realising I didn’t know how to program. Not just that, but I assumed I simply lacked the skills to figure it out, and didn’t have the slightest idea where to start. A few attempts to create some form of tabletop game went (understandably) nowhere, and I decided it would simply never happen.

Obviously, I was wrong, otherwise this blog wouldn’t exist, and I wouldn’t now have a game rapidly ballooning into multiple Python docs.

2) I like a challenge.

I decided to stop treating programming as something I could never learn, and instead, simply as a challenge, the same as anything else. For the time being, therefore, until I successfully manage to spread word of U.R.R, my ‘motivation’ is simply the challenge of designing the game exactly how I want it, and figuring out the code to make that happen.

3) I think other people like challenges.

Completing a game of NetHack
Completing NetHack; absurdly hard, but fair, balanced, and damned satisfying.

Now, this is not to walk into the debate on ‘what makes a game’ – is the game about the challenge (e.g. Super Meat Boy), the complexity of gameplay (e.g. NetHack), the story-telling (e.g. Portal), or how close it can get to cinema, the closest comparable medium (e.g. Heavy Rain, Mass Effect)? This will be discussion for another time, but succinctly, I think the term ‘game’ for everything we now call a game is much, much too broad. The first two on that list and the latter two are two completely different creatures. I, however, aim to emulate the first two, with a little bit of the third. I want to make a game which is challenging, and tricky, and requires a lot of thought to get through, because that is the kind of game I enjoy the most and derive the most satisfaction from completing.

I think Super Meat Boy and NetHack are supreme examples of what games can be due to their challenge, though one relies on split-second reflexes (and, partly, memorisation) while the other relies on thinking on multiple levels and devising long-term strategy to deal with the many thousands of different threats to the player. I might have thought Mass Effect 2’s storytelling and immersion were impressive, but I didn’t feel like I’d accomplished much because it just wasn’t that difficult. Thus, reason #3: I want to create a challenge people will enjoy tackling, and gain satisfaction from overcoming the complicated trials of logic and strategy I’ve devised. I want to give people the feeling of pausing, taking stock, devising the best strategy, and then seeing the results of it.

A key part of this is consistency in the game universe – which is to say, you should be able to figure out what (for example) throwing potion x at monster y will do, very specifically, even if you’ve never done it before. What exactly I mean by this will be covered more in a later entry.

4) A gap in the market.

There are lots of roguelike games out there, even if those who play them are focused around the big ones – NetHack, ADOM, Angband & its variants being likely the biggest three.

However, with the exception of rare allies, you are a lone warrior in all of these games, and a warrior with a clear objective – retrieve the Amulet of Yendor, save the world from the forces of Chaos and slay Morgoth, respectively. However, a ‘strategy roguelike’ it almost non-existent. While the player will begin as a lone warrior before you get anyone behind you – a task harder than simply saying to someone “Will you join me?”, a la Fallout 3 – you’ll be able to get groups of almost any size following you. You won’t have direct command over them (you can delegate tasks, but the AI will decide on its own the best way to carry it out) but you will still end up with conflicts between very large forces on both sides, rather than having to make your way through a vast world all by yourself.

5) Emergence is cool.

Ultima Ratio Regum roguelike stalemate.
A temporary stalemate.

As this moment in time, the enemy AI in U.R.R. is roughly at the level it will be when I release a demo. The ally AI is coming along, but needs some work on behaving correctly around you and judging whether you are being led – which is to say, the creature is stronger than you, and considers you subservient – or whether you are the leader of your little group. However, despite this, spawning numbers of creatures on two different ‘sides’ has produced a bunch of behaviour that I had no idea they’d do. This has included a point where the two AI teams ended up in a stand-off; neither felt they had the mass of units to assault the other successfully, but were also sufficiently confident to not simply flee altogether. On another occasion, enemies who had fled from me had just rallied themselves and regained their confidence, turned around to face me again – only to see me slay another one of their allies, and instantly shatter their fragile confidence once more.

AI is one of the biggest focuses in Ultima Ratio Regum. It’ll be a lot more advanced than it currently is before I release a demo, but even now, it ties perfectly into the goal of the game. There will be no fixed objectives, and no missions, and certainly no quests – rather, a large and complex world you can interact with (read: mess with) however you so desire, and one that will respond and react to whatever you do.

So, there you go. Those are the big five reasons. Feel free to post any thoughts about my logic, goals in the game, or what you think the role should be of ‘challenges’ in games.

Coming Saturday: A release schedule, randomisation, and identification.

Coming Tuesday: A first look at the AI. Flowchart included!

In the mean time – I’d check out the About U.R.R. page.