Breeding unpredictability...

BREEDING UNPREDICTABILITY:
The above deck is just a "cover" and unrelated to this article. It was chosen for this by looking rather cool.

I use evolution to design decks for me, and part of the process is to breed the decks under the harshest conditions possible.

Part of the conditions is giving the opponent omniscience. They get to look at my hand whenever they want to, which gives them a pretty big advantage. For example my opponent can see wether or not it pays of to cast a thoughtseize or even a counterspell. They can also plan out their attack better if they see no removal.

The omniscience rule was originally established to simulate games against an opponent that is good at reading body language. Someone like that will be able to gain a feel of your hand as the game progresses, and the idea has always been to hope that evolution could somehow find a way through that.

Today we agreed that there are enough signs of this taking place regularly. The "signs" in this case is having an opening hand that makes my opponents adapt a playstyle or make choices that are regretted a few turns later.

The project has barely begun in terms of generations and yet the decks that we are evolving are showing signs of being harder to read. They contain just enough unpredictable card choices to make my opponents react badly to being omniscient.

In other words, the decks are evolving a resistance towards discard for instance, not just by including cards I want to see discarded but by containing cards in certain amounts that they are drawn after the initial discard has happened.

This was always just a theoretical part of the project, but seeing the deck topdeck the right answers through enough games is actually rather awesome.

So far this hasn't been enough to turn the games around, but that is basically just a matter of time.

The larger question "what is the limit of adaptions" remains unanswered. The many pressure points that I use to breed resilience into the project hasn't broken it so far. On the contrary the project has taken on a life of it's own. The aspects we wanted to breed into the decks used to be directed by me and my lead tester, but once it was discovered that some decks were adapting features that allowed the project to basically change it's own dna I had to rethink our procedures.

To give an example:
During the evolving of two halfdecks (green and blue) the original theme was to go for a green blue combo based on sprouting vines. Evolution disregarded that plan and chose to focus on land grant and lay of the land instead, and opted for a low amount of forests that got thinned out. Since then we've fed evolution with cards that thin the deck and evolution is loving it. Most recently merchant scroll has been added to the project, and this is where the whole thing gets interesting.

You see merchant scroll becomes a way for the blue deckhalf to manipulate the development of other halfdecks. During what is called "the mutation phase" merchant scroll have a chance to alter the "dna" of other decks.

An example is in order.

Let's say that the blue halfdeck gets coupled with a red deck with absolutely no blue cards and we got a situation where casting merchant scroll isn't helping us if we search for anything within the blue half, how can this lead towards drawing something useful from the red half ?

The answer is fire // ice.

This way merchant scroll can become a driver of mutations within other deck halves, and those mutations tend to bring other mutations with them.

If at a later point the red deck has gained fire // ice as a mutation, then that card becomes a driver for adding cards that interact with blue mana. If the deck is developing towards using lotus petal for example the deck suddenly has access to casting the ice half of the split spell.

At the moment fire // ice is the only common split card in premodern, but this has been an example of how drivers work.

The discovery that evolution was capeable of having one half manipulate the mutations of another half (and so manipulate the overall project) came as a shock to our group, and we've been working at implementing this aspect into the project whenever it's been possible.

In my original project this was only discovered after evolution has bridged several themes together in the halfdecks before we discovered what was really going on.

The full scope is to feed evolution with any powerful drivers we discover during the forming of the halfdecks we currently got.

Thinning is currently our primary "first driver" but other themes like flashback are popping up. It's impossible for us to predict the directions that any of the drivers will take the project, but that's sort of the point of it.

The brewer in our project is evolution, and we are just there for the ride.

There is basically nothing new about those "drivers" in the golden days where meta stagnation was an unknown thing it was the use of "shells" that was the driving force of innovation. A shell is basically a small group of cards, that put together protects your deck from a certain theme. Counterspells and discard was among the earliest form of shells, but the core of the concept was that shells could fit into basically any deck that ran the colors of a shell. Once shells were discovered as a powerfull tool the number of shells exploded and created a wide variety of concepts. Shells basically died when statistically minded people (like me) figured out that you could build a deck out of the most successfull shells. This was the first real cause of stagnation. People went from innovation to just recycling every old concept ever discovered, so drivers is far from being a new concept. What might be new, is that evolution can tap into the exact same concept.

However, the discovery that evolution is actually finding ways to counteract omniscience is completely mind blowing...

• article
• evolution