Wheel To Wheel: Design Diary
I've spent at least three years on the side working on one game, and for most of those years the entire game was only eighteen cards. I've rotated every permutation of blackjack around my mind that I could think of, before setting the idea down for months at a time. This was meant to be the first analog design I would finish and would actually be the fourth. Unlike other microgame design diaries this will be much longer and proceed chronologically. If you're interested in playing, the print & play files are available for free here.
Before The Microgame
I started by wanting to make a racing game. I've been a pretty longtime fan of Formula 1 and MotoGP, and had played a lot of Formula D. Later I would play other good racing games like Heat: Pedal to the Medal and Lacorsa, but starting out I wanted to take my own spin at things and make a small analog game of my own.
I definitely wanted some kind of bidding or push-your-luck element. The now giant folder in my notes program is named 'pylracer', and this basically reflects where the project was anchored. This is not an original idea of course, Formula D operates very well as a fun dice-chucking push-your-luck game. I wanted something card based and more focused on close interaction between players and less the players versus the track itself.
What came out of this initially was frankly very strange. Players had their own decks of speed cards, and the track was a perfectly circular grid of spaces that they moved around, endlessly challenging each other on 'corners'. I honestly don't remember exactly how it worked, but it got far enough that I printed out some physical versions to try.
What eventually sank this whole operation was primarily not really understanding how to get what I wanted, on top of not having any real constraints. I decided to set it down and make some microgames instead.
A Million Backjack Variants
In the background of making three other microgames I would noodle on this one. With only eighteen cards (and two players instead of up to four) came a lot of blackjack variants. They all shared a similar structure: a sequence of corners and a small deck for players to pull cards from to try and win a corner. Sometimes the value of the cards mattered, or you were looking for certain symbols, or adjusting the bust threshold of the corner. It was all basically blackjack.
One solid idea fell out of all this experimenting: it was more interesting to follow how many cards a player pulled from the deck rather than the specific value. Imagine a deck of values 1..12, or 2x 2..6. You want the player to pull at least three to five cards from the deck so that there's enough decision space each round to feel interesting. But with that many cards pulled, the variance of the total value drawn is extremely high, anywhere from single digits to over thirty. A player can't control what cards they get, so instead it becomes much more interesting to evaluate how many cards they pulled. This has the very fascinating effect of saying: this player will win the round because they have more cards than the other player... unless they have busted!
This was a real breakthrough, because it felt like racing! You can imagine yourself in the cockpit of a car or astride a motorcycle and see your rival ahead of you, they're winning! Unless they overcommitted and are about to spin out or go wide. Should you pull another card or not?
Decision Space
Unfortunately while this idea proved powerful, it didn't immediately seal up a design. High number variance still proved an issue, and any attempts to compress or remove the numbers naturally drained any real interest in the cards being pulled. Many ideas felt like one player was predestined to win the corner, regardless of what either player did, or a player immediately busting on their very first draw from the deck.
Another breakthrough (probably a year later now) felt much more interesting: a shared bust. When players pulled cards, the value pulled added to a shared bust value set by the corner. This felt like it neatly solved the variance and predestination issues, as cards your rival draws count for busting just like your cards. At first it was promising, your rival pulling a high card from the deck meant something for you both. This sort of tied together nature also feels weirdly appropriate to racing - many previous versions tried to simulate crashing, slipstreaming, inside/outside or other little dynamics; this idea feels like you're sharing a space with someone else.
That said, this still wasn't enough. Shared busts worked quite well the first time, but no one who played felt a strong desire to play again or unpack how they could have played better. The system functioned but didn't provide much room for the player to make interesting decisions or express themselves. Functionally the only thing they can do is pull another card or stop pulling, and turns out the predestination problem didn't actually go away!
The Final Version
A few quick variants and ideas were stepping stones to the final idea that hit me like a lightning bolt. Having made four of these microgames (and many other games besides), the good ideas really do hit you. Not that they come to your mind instantly fully formed, I mean obviously you've just read many paragraphs disproving that, but they give you a jolt and just feel right.
The idea here took everything before and played with the one element of the design that stayed in my blind spot this entire time: letting the players adjust the deck composition themselves. Now what happens is each player gets five cards from the deck and can order them any way they want. Then they reveal cards one by one, adding to a shared bust. There's some hidden information in what your rival is doing, but you know exactly what cards you'll draw. What you don't know is if that card you'll draw will put you over the shared bust threshold. You're still playing to a number of draws, so it's clear who will the corner... if you don't bust.
It feels like such a small idea but it exponentially expands the player decision space and gives them so many things to try. A hand of five cards can be a nice even spread of high to low, or all high or low, or any weird mixture in between. All the combinations are playable because you're now fundamentally pitting your hand against what your rival will assemble. Many previous variants were easily spoiled by getting a higher card too early, now a player can say "my third card will be the 6" to try and intentionally bust and trap their rival into drawing more. There's a crucial zero in the deck, which is a "free" draw, but I have myself literally lost a corner by drawing it looking for an easy win without realizing we had busted.
Now the game truly feels like racing! So many early versions tried to play with number progressions or whatever to feel 'fast', but this version does it right. The deck you assemble does feel like how racers attack a corner: braking late or early, feinting or making late lunge. Playing high cards early feels like carrying too much speed, timing a low value card feels like knowing exactly where you'll hit an apex.
One last tweak was made to add the ? card, which is an alias for a value printed on the previous corner. This gives the tracks some overall identity instead of just a collection of challenges, and makes a previously fast corner feel fast as you're going to the next corner. It also adds a second 0 value to the first corner which I feel is appropriate for how races start.
Takeaways
Main thing learned: it's incredibly hard to meaningfully tweak blackjack. This was pretty obvious from the start and remained true throughout. Any game that has a meta-structure of gambling can skate by without the complexity needed to remain interesting. Hell, a lot of gambling games don't even have any decisions at all. I kept thinking that with enough small decisions over time I could overcome that, but it never happened and probably never would have.
Individual ideas can feel good, but over the course of the design I could see when an idea didn't have the juice to do what I needed it to. In the past three years I generated probably around forty configurations or evolutions of the game. Most of those I never even tried, just wrote them down so I could forget them. A smaller (but still large) portion got tried on friends but frequently wouldn't stand up to a whole game being played. Only a handful got serious consideration.
The jolt of a good idea I described earlier is a great feeling because the idea might contain a delightfully unsolvable problem that is at the core of the best games. I'm drafting a blog on how I see those kinds of unsolvable problems, but it essentially means there is either a very large or infinite decision space that the game provides. All but maybe one or two of the design iterations tried in this project had a decision space that was way too small, typically in the realm of "draw or stop drawing". This works for blackjack, but I couldn't patch together something to replace the gambling meta-structure. The final idea not only taps in to "what cards does my opponent have" but far more importantly: what did they do with the cards they might have? This will likely be the biggest lesson I learned.
At the end of the day I am weirdly proud of how long it took to get to the finish line on this project, it validates my feeling that it was possible. For all but the final few weeks of this process I wasn't sure how it would manifest but I had a feeling it would. There isn't a better way to learn game design than to put in the practice, learn from failures, and build an intuition for how games work. Three years and forty versions later I have a game I am extremely happy with and a very deep understanding of how and why it works.