#FAQs

#Why Does It Keep Turning Grey And Reseeding The Photobots?

When Photobots are first placed into the picture, they can be placed as adults or newborns. If they are placed as newborns they come with a nominal amount of yolk, an energy source they can draw down on before they learn how to feed themselves.

Assuming you have left auto-evolve turned on, the population is spawned and then needs to learn how to eat, move, mate, avoid dangerous contact, learn what to eat and what not to eat, and learn which coloured bots make better mates. That is happening even though the bots themselves keep changing colour depending on what they have just eaten and how fast their metabolism is. It is a lot. What usually happens from scratch is that the bots twitch a bit and then die.

Here is where forced evolution comes in. You see the picture shift slightly and turn grey. Under the hood, a fitness test is being applied to each bot from the iteration just past. The system judges age among peers, to a point. It also judges reproductive ability. We want more bots that can breed. Ideally we want bots that can move, but we also want to protect the possibility that a stationary, or very slow moving, creature might emerge to solve the same problem another way.

To stay alive, bots have to solve quite a complex problem. They need the right sensors. They need the ability to move to food, attract food, opportunistically consume passing food, or just get lucky and spawn into an energy-dense pocket they can eat. Eating has consequences. Right now colour affects food preference, mate preference, digestion, waste, and the visible ecology.

A system external to the main simulation picks the best of the best bots, breeds them by genetic crossover, John Holland style, and pads out their numbers with strong material from previous iterations plus a fresh batch of random bots to keep things noisy.

The bots have 558 genes. Each bot genome is expressed into an embodied form within the simulation. Founder bots are placed into the energy map derived from your photo. Placing them draws embodied launch energy from the world, so even the act of reseeding slightly changes the original image.

That is a lot of computation. Filtering, selecting, breeding, reloading the world, and placing the next seed population can take from a few seconds to a few minutes, depending on the run settings, the number of founders, and your machine.

#What Is This Energy Map Derived From My Photo?

When you choose a local picture, your browser prepares it on your device and derives a kind of universal world energy map from its colours. The selected image is not uploaded to a Photobots server. The energy is measured in phWatt, and it is what the bots eat. More detailed image-use guidance is being prepared and will be published only after specialist review. Separately curated gallery works have their own clearance record.

It is as if the universe has been reduced to one fundamental force. Bots react differently to different flavours of energy. They digest it into different reservoirs inside themselves, and those reservoirs drive muscle actuation, thought-like control work, repair, growth, mating, pregnancy, and the rest of their tiny lives.

Everything a bot does is inefficient. Waste energy accumulates in a waste store and is expelled when the bot's body (as expressed by its genome) says it should be. That waste energy is absorbed back into the world energy field and projected back as colour on the screen.

Think of it like a hologram: the field, including the bots living inside it, is the simulation's real state, and the picture you see is the visible projection.

#How Long Will It Take Before It Stops Flashing Grey?

For it to stop flashing grey, the bots need to evolve to the point that they not only live a lot longer, but breed in sustainable numbers so the population never drops below the configured population floor.

In my own testing, including long development runs, I have seen movement towards more sustainable bot populations, but I have not yet seen solid evidence of a stable self-sustaining population. We are asking a hell of a lot from evolution right now. Nature had billions of years and an absurd number of chances; Photobots is trying to do something interesting in a browser while you watch.

The development CLI can already run long trials and export diagnostics. A more public-friendly packaged CLI, stronger public evidence, and easier archetype sharing are still being rounded out. If you are running from scratch in the browser, expect it to keep flashing every few minutes, even after many thousands of iterations.

#So What's The Point Of It?

Ultimately this project is being built towards a physical installation, where powerful computers could drive individual displays running high-resolution images with very large populations of bots for months, years, or permanently. That is the big goal, not what I expect any laptop browser can do today.

That said, one beauty of it is that there is simply no way to tell what the bots will evolve to do with your pictures. In time, collections of successful archetypes will be available through this or related sites. In the browser today you can save PNGs from paused runs, inspect bots, export diagnostics, and work with saved lineage or archetype JSON. The friendlier public exchange of evolved lineages is still coming, however.

So that is one point: art.

The other is stranger. Photobots represents a collection of mechanics, each of which manifests inside the bot and the world of the bots in ways that echo contemporary theories of mind. The bots evolve through genetics that express bodies with complex, inspectable controller networks. These minds change over the lifetime of the bot through feedback within themselves and through contact with the real, image-derived world.

Philosophical ideas such as "no free won't", anxiety as a genetically expressed mechanism for sensing backpressure from real events on projected desires, or tuple spaces, mix in with each other to form inspectable creatures whose awareness-like state, attention, intents, and actions can all be examined.

With all these mechanisms and this apparent inner life, do these creatures cross the Chalmers wall? Or is Dennett right that there is no wall to cross, only mechanisms becoming rich enough that our language changes? What if we layered on more mechanics? Or is it enough that we can now ask, more sharply, "what is it like to be a bot?"

#How Should I, As A Novice User, Use Photobots?

For now, just play with it in the browser. Explore the options. Let it run for a while. If you like what you see, pause it and save a PNG. If you have saved lineage or archetype JSON, try it on another image and see whether those bots can make sense of a new world.

Have fun, but be patient. You will usually have more fun with bots that come from a line with many ancestors than with brand new random founders. Soon enough I will add more of those here, along with an expanded gallery.