Creatures Caves | Community

An article by Andrew Carroll on the potential of using variables to create COBs with genetics.

Its been no secret that Andrew Carroll has believed this is a way forward for some time, but on the CyberLife side things have been quiet since one appeared. Having taken a look at all the games Andrew now expresses his views on genetic objects and their importance in the follow up to Biology Vs. Creatures.

Genetic Objects, Why Bother
The first genetic object was released by Cyberlife about a year ago. It was the dark green cactus, and it had a simple little ov (an ov is a cob trait or variable) that controlled the rate of its timer tick. Since then, there has been relatively little progress in genetic objects. With the sheer mass of cobs currently available, and the limit on active scripts, it seems that the new goal in cobbing should be quality, not quantity. In addition, as more and more room cobs are made that change the layout and attributes of Albia it is becoming more important for cobs that rely on environment conditions to be more adaptive. Add that to the ability to add new features that the cob itself can develop, and you can see why genetic objects are so important.

The principles behind genetic objects are the same principles which drive genetic algorithms. It is the idea that by employing small mutations in the code of a program or object, that code will be able to write itself. By employing a very small piece of mutable code, it is possible for a cob to do many more things than a cob written with no mutable components, and for much less scriptorium space. Take this scenario:

You've just made a great cob, it relies on light, organic nutrients, inorganic nutrient, and temperature to determine where it can grow. The goal is to get the plant to grow in an area where there will be little other life to provide food in a food poor area, but to keep it from growing where it is not wanted. You quickly discover, to your great disappointment, that these levels are not set in stone. As soon as a good sized population of your plant gets going, one of the levels the plant depends on changes just enough to kill them off. I think this is one of the problems CL has had in maintaining stable ecosystems in C2 and C3. In any case, if you use small genetic markers you can keep the population from dying off when the season moves from spring to summer.

It's a nice thought, but how do you do it? Well, that too is very simple. Give the object some object variables (ov's) that will be used in that objects actions. One idea is to tie ov's into stim values, so that norns can get a balanced diet from the same type of objects if they have genetic codes. Another is to mutate the allowable level of heat, light, or other environment conditions to allow adaptation to changes in the environment. When an object "reproduces" have it set it's ov to a va and use a small random number to change that va slightly and pass it onto the new object as a ov. The smaller the amount of random change, the slower evolution occurs, but the more effective it is after an object has evolved to an ideal state.

When done properly, genetic objects can be vastly superior to those lacking them. They can be used to balance an ecosystem, evolving to adapt to unexpected levels of a predator or prey (something that I think would have greatly improved C3). Giving small genetic ov's to norns themselves might allow blood types to be simulated. It might be hard to do properly, and might require additional testing to get everything balanced right, but it allows features in a cob that can be very useful.

By incorporating genetics into cobs, you tap into a small part of the system that brought small globs of cytoplasm and nucleic acids to develop into many collective globs of cooperating cytoplasm and nucleic acid capable of generating intelligent thought. Using genetic traits you can get cobs to do more things with greater flexibility and less work. Why would you not want that?

Originally published on HomeCreatures.