Post by: Joshua Buergel and Grant Rodiek
This is the latest post in our weekly series about developing Hocus Poker. If you want to try Hocus Poker, you can find the PNP here. Or, read the rules here.
This week, we want to discuss the sets that players can create in Hocus Poker. By sets, we mean things like a Full House or Two Pair. Just like in normal poker, your goal in every round is to build the best set. Some sets are better than others, and as we narrowed down our list of sets, we found some choices were better than others.
This is a controversial topic and everyone has immediate expectations when they sit down to play a game with Poker in the name. We wanted to take a moment to prove we aren’t nuts and that we’ve made thoughtful choices to choose the right set list.
Josh: Obviously, at the core of Hocus Poker, is, uh, poker. It’s kind of right there in the name. Or, at least, it is now. If you asked people to say what poker is about, they will probably list a few things: betting, bluffing, and the poker hands. One of the earliest things that we tinkered with was exactly which of the poker hands we were going to have in the game.
One of the first questions that Grant asked me, when I came on board originally as a tester, was if I thought that it made sense to expand the number of hands in the game. Being a fan of strange stuff happening in card games, I thought it sounded great. One of the “hands” that comes up in poker is three pairs (ex: two 6s, two 3s, two Kings), which is always good for mocking your friends who have it. That one went in early, along with five of a kind, which was reachable sometimes due to spells.
I was immediately quite curious what the actual odds were for three pairs, which I needed to know in order to slot it into the proper place in the ranking. Being that I’m a software developer, I decided to just write a simulator for the thing. I’ve done that a fair bit in the past for other games, and have always found it to be quite handy.
I could have computed the odds mathematically, of course, but the simulator is more fun to write. It ended up being a tool that we used quite a bit while working on the game. Three pairs was initially slotted below straight, but that turned out to be too low for that hand. We also added double threes (two threes of a kind) at this stage of the game.
Grant: Three Pairs, Five of a Kind, and Double Three (two sets of three of a kind) were all pretty intuitive as far as creating the hand. They also happened with some frequency and everyone always asked about them. Our players had the same gut instinct as Josh — this is already a goofy game, so why not?
This, ultimately, brings us to one of the core issues we’ve needed to deal with for every change in the game: how intuitive is this for players?
The answer was, not very.
For people who have played poker, which is a large number, we had issues when we changed the core set list. This, by the way, is:
- Straight Flush
- 4 of a Kind
- Full House
- 3 of a Kind
- 2 Pair
- High Card
Inserting a three pair in that list means we suddenly have more hands and things don’t mean the same. It’s like changing the meaning of a single word in a language and expecting people to just roll with it. We learned really quickly that we needed to be really thoughtful about how we altered these hands.
This forced us to take a step back and ask: If we’re going to change the hand list, for what reason? What justifies the change?
Josh: The tell (to use a poker term) is that even long-time poker players were constantly looking at the hand ranking list to see where the new hands fell. The upside was novelty, seeing fun stuff happen that you don’t see in poker. But the cost was significant, in terms of down time in the game as people scratched their heads and tried to figure out what they had and what they could make.
Was the novelty adding much? To my surprise, it seemed like the additional poker hands were worse for poker players and not novices. Novices were having to look everything up anyway, so another entry or two on the list wasn’t really hurting anything. But even just a few new entries on the list was almost reducing everybody back to a novice state, at least for a little while. As a consequence, due to that playability concern, we backed off of the goofy hands and went with a straight set of poker hands for a while.
However, there was another issue that was coming up for both of us: how common flush and especially full house were.
Grant: This problem became so bad at one point that if you didn’t have at least a Straight, you should fold. It just wasn’t right. It reduced our eight hand game to a three hand game.
When we started to solve this problem, the approximate setup for each round was 4 cards in the Square, which is our term for community. All players had 2 cards in hand by default. The two spells in EVERY round let you add cards to your hand OR add cards to the square. This meant players could have 8, 9, even 10 cards with which to make a hand.
We had some really crazy solutions to this, including changing the ranking of the hands (terribly non-intuitive), making it so that the same hand couldn’t win sequential rounds (band aid fix, adds new rule), or just cutting out lots of hands (non-intuitive, ruined our nice distribution of hands).
With time, the right solution was the simplest one — we decreased the number of default cards in the Square from 4 to 3. That one solution had a massive impact and largely restored the distribution in most cases, except for the Full House.
Now that we isolated the issue to the Full House specifically, we took the content route, as opposed to the system’s route, to fix it.
The fundamental issue was that Full House occurred too often for its relative strength.
Josh: If you remember, there was actually a period of time when we had 4 cards in the square and 3 in hand. Add in the extra cards from spells, and a full house was almost a minimum to compete.
Before we get to the full house problem, the next big change for hands was thinking about naming. At the time, we were looking at the possibility of pitching to publishers. We’ve mentioned that thought experiment a few times, and it was mostly good, but one of the blind alleys for us was renaming the hands. The thought behind it was that it would give a little bit more theme in the game, it would further distance our game from poker to help manage expectations, and for novice players, it wouldn’t matter too much if things were renamed. We figured that most poker players would probably just keep referring to the hands by their poker names anyway.
Here, again, our expectations weren’t really met. There were references to the hands in various places on the rules and cards, and people were having to unpack that reference each time we saw it. It was an unnecessary cognitive load on our players.
Grant: Just as a general lesson, never force people to re-learn things that are so common. This goes for a lot of things, and is a lesson for pretty much any form of design. For example, a video game developer is foolish if they mess up the super established UI framework of World of Warcraft when making an MMO. Or the FPS developer who tries to reinvent the console control scheme made standard by Halo.
In card and board games, if you have DOMINANT reference points like Dominion, Magic, and Poker, you want to tread very carefully when shifting standards that are just so well known by so many players.
Josh: It was around this time that we also started getting antsy again about the menu of hands. If we’re trying to distance ourselves from poker, the thought went, why not go all the way! Here’s where my simulator came into play again. I could pretty easily plug in a bunch of different hands and try them out. Why not? We could see where things fell in the spectrum and see what might shake out. Grant and I put on our brainstorming hats (Carmen Miranda fruit hats) and decided to see what we could cook up.
This was also the period of time when we were flirting with five suits.
Here’s the list of hands that were tested in the simulator:
- Rainbow (one card of each of the five suits)
- Three pairs (again)
- Double triple (again)
- Two pairs, but of adjacent ranks
- Four card straight
- Four card flush
- Skip straight (every other rank), in versions of odd, even, or both
- All cards ranked below six
- Three and four card straight flushes
Some of these were clearly terrible, but testing them in simulation was cheap, so why not? Some of them were motivated by an idea that we might be able to transition things to a game where you really only care about straights and flushes, which might be an interesting twist. Due to a bug in my code, it looked for a while like Rainbow was going to slot in at an interesting place, so we tried that one out on the table. Grant played one game with it, and it was ridiculous, which led me to looking more carefully at my code. In the end, most of that junk just wasn’t worth the cost on the brains of our players.
At this point, we basically had the poker hands, minus full house. It was just still a little bit too common, too easy to go for. I want to say that we had also dropped something on the low end.
Grant: We dropped high card as a hand because it never happened and it’s actually not that intuitive. For a while I was really on a warpath about using only intuitive hands. Many of them actually are, except for full house, which is a bit strange.
This was one of those weird phases where we were both scratching our heads and most of our ideas were just bad. For a while I think we both worried about being overly negative — I know I did. Josh would send out a new report from the simulator and even when the probabilities seemed right, some of them just didn’t make sense. Or, they were too clunky.
I especially hated the four card hands. It just felt like a lame, Diet Coke version of the real hand.
Some of the other ones, like “the summed strength of your hand can’t exceed a certain amount” were interesting and may one day become expansions or variants. But, we still had this huge gap in probability between flush and four of a kind. A few times I suggested we just cut straight flush, maybe even four of a kind. Josh, rightfully, thought this foolish.
Many of these ideas never left our emails. For a brief phase, we just tested without a Full House in the set list.
Josh: We did keep getting tester push back about the lack of Full House in the list, though. Players who knew poker wanted it back in there. And, there was a gap in the probabilities as well. The leap from flush to four of a kind was a substantial one, and it would be useful to have something to fill in the gap. Just not Full House – it had proven itself as just a little too easy to get.
We strapped back on the thinking helmets and started firing ideas. What about the four-card straight flush? Would that be confusing? Anything else come to mind? I was standing around at a party holding a beer when I started just visualizing poker hands.
I was thinking that they’re composed of three basic elements: straights, flushes, and sets of same-ranked cards. We had a hand that combined straights and flushes (the, uh, straight flush). What about combining the other ones? Combining a flush with a set doesn’t make much sense. I guess you could do something like two pairs, which are also from the same two suits. But that seemed odd. Straight and set, though, that seemed promising. That’s when I thought about a three of a kind crossed with a three card straight.
As I originally conceived it, the three of a kind needed to be the middle card of the straight, which makes your hand a cross when you lay out the cards. I plugged it into the simulation and, sure enough, it dropped right into the right spot in our probabilities. It then seemed obvious why: this was really just a Full House, but with fewer probabilities for the other two cards. When I proposed the hand, Grant suggested that we loosen it up so that any of the possible straights using the three of a kind would work, and that’s what stuck. It’s about half as common as a Full House, but still much easier to get than a four of a kind, so it slots in really nicely. I also like the way it adds just one unique hand to the list, just to give us that twist.
Grant: I love getting a crossways, so much that every time it happens I send Josh a picture. There’s something special in that it’s a hand that’s Hocus Poker’s. We’re very protective of it. We’re going to patent troll the stuffing out of it.
We did a lot of experimentation, but at the end of the day, the solutions were quite simple.
- Four suits, not five.
- Three cards in the square, not four.
- Two cards in the hand.
- Crossways instead of Full House.
- The rest of the hands? Well, leave them as they’ve been forever. Don’t change what isn’t broken.
This exercise was a classic case of experimentation to confirm that things don’t really need to change. Sometimes you don’t need a massive change or complete overhaul. You just need a tweak. Our contributions in changing the game of poker came mostly in the form of removing player elimination, creating a market driven spell activation mechanic, and introducing spells in the first place. The sets? They worked (mostly) just fine.
If you want to try Hocus Poker, you can find the PNP here. Or, read the rules here.