The Project

(Originally published August 13, 2009)

The first days of The Field of Blobs tend to drag. This has been noted. Evolution takes some time; millions of years usually. Even when the life cycles are clock cycles of a multiple gigahertz computer, it’ll be a bit before anything challenging crosses your path.

In the mean time, you can make your own blobs. The player draws up some simple plans, blueprints, prototypes or whatever. Those plans are put through maybe a GEP algorithm to find the closest match to what’s available in the possible genomes preset in the blobs. I can’t describe how incredibly vast we’ll have to make the basic genome of the blobs to make it as diverse as possible, as to consistently evolve from chaos into something random yet still designed. When birthed, the player blobs can be pitted against each other or even released into the wild. The domestic blobs have an advantage against the feral blobs at first due to the artificially enhanced evolution rate and designed purpose. But soon their genes will spread into the feral gene pool and become another source of diversity, quickening the evolution and sure death of all players in the field.

Those are not set free are battled against each other. Mostly it’ll be a rip-off of Pokémon. But I guess in any game that allows breeding also has stupid tournament fighting. Sigh. God I hate conventional things. I’ll do I can to keep them out of these games but often they’re still expected.

I’d like to see the choice made by the GEP algorithms that any blob not easily survivable with player give tools and abilities be survivable in different ways. If they’re made fuzzy and cute, they’ll have to breed like crazy and have super short gestation time and turnaround. If made with certain advantages and no flaws, they’ll have to have other weaknesses. How do porcupines breed? I’ve no idea. But it’s something the player will have to think about when making the sea-urchin blob. Those are the challenges you’ll be faced with.

Maybe there will be a limit to the cycles given towards evolutions in the artificial evolution pit. There will be a severe difference between the growth mutations of asexual and sexual blobs. It’ll be another choice to make when designing your blob. Also levels of care should be considered. You can breed in traits/temperaments to your blob: Aggressiveness, skittishness, isolationism, pack hunter, loyalty to their own, loyalty to you, etc.

The Trait Problem: I have no idea how to have the base traits like these already present but not active in the basic blob. This game is supposed to be the simplest concept of ours and it’s soon becoming the more and more unimaginable to program. Oddly it’s following the evolving patterns of life; started simple but becoming exponentially more complicated. I read an article about retaining learned behavior in Video Game AI while promoting further need to learn. This could possibly solve the trait problem but it’d have to be a sublevel of memory, genetic memory, or even an analogue for instinct.

I’m still not very adept at programming but there will have to be two levels of learned behavior. One will be the Instinct Layer and the other will be the Cognizant Layer. The IL will have rudimentary skills all life eventually develops. Why deer run from everything. Why rabbits eat what they eat. Why beavers mate for life. This will all be captured, stored, and expressed at this layer. The entire genealogical history for this strain of blobs will be here. The trick will be making access to the storage somewhat muddled. I don’t remember when my monkey ancestor got mauled by a jungle cat. I just know to be scared of them.

The Physics Problem: How can you inspire real world genetic solutions without a near perfect physics engine. We got our legs from needing to move and reach things only a biped could. It can’t be a simple value hiding behind a generic choice. The choice between a movement-speed of 1 vs. 2 will always 2. We can’t have a large selection of choices in locomotion. Three bits of choices, bipedal, back-jointed bipedal, six legged, eight legged, multi-legged, wheels, slither, ooze, and whatever else nature does, cannot be called random or evolved. Those are just data sets already available. Even with a clever rule-set–pros and cons for each—can I accept this over genuine evolution, which is why I’m so enthusiastic about this project. The whole point is to not know what conclusion will be drawn. We must provide the problem of locomotion and let GP solve it uniquely.

In my mind, I picture a blob with a grown tentacle evolved for grabbing food missing and hitting the ground cause it to lurch quickly in a certain way. This leads to purposeful whipping for better movement. Others will learn to roll. Others may even learn a complex way of undulating their blob bodies. Maybe a top-of-the-line physics engine could handle what’s necessary, maybe. As an independent company, we won’t have one of these. Maybe when we start up development, we’ll have a moderately capable one that can at least lend to some quality physics and provided a good enough backdrop for artificial evolution.

Note: This is not a problem for Alphabet Soup. Simple physics and their starting shape should work gloriously in spurning interesting evolution.