Hot take: People dislike anthropomorphizing vocabulary when talking about LLMs. Words like “think”, “dream”, maybe even “you” when talking to an “it”. However due to how the beast works, it is actually immensely useful to refer to it as a human, not because it is, but because on the vernacular it was trained on. However it should be possible to re-train a model that will completely separate how humans function from LLMs. That would open some interesting options.

It took me a while to form a solid opinion on AI. Then I realized that it will probably shift again, so it would be good to write down my current thoughts so I can compare them to whatever comes later.

I’ve used

Wrong Goals

Middleware

Failures

There is no shortage of stories of how AI has failed to do this and that.

Disdain

This is an interesting one, and I observed this a lot myself. This failure comes from inherent disdain for the

History was not lost, but much was forgotten. If our records are correct, what we call the “age of prosperity” has ended over 12 generations ago.

Our civilization was bold and hopeful. We have perfectioned perpetually self-sustaining habitats in the frozen poles of our planed, and have finally obtained the technology necessary to colonize space. Initially, we started building two massive habitats in the orbit. We were also staged a launch of a fleet of ships on a journey to the nearest planet. Although barren, it had similar gravity and a core live enough to support a magnetospehere.

Economics of colonizing the orbit did not pan out. Radiation and micro meteorites turned out to be a much bigger issue on the long term than expected. Project was soon reduced to a single habitat, and eventually that got downsized to a single orbital ring, mostly for scientific purposes.

The majority of our fleet has reached the promised planet. First generation of its inhabitants was selected people who were explorers, scientist and heroes—people willing to make a sacrifice to obtain a legacy. They were hard at work building a self-sustaining city, but were tethered to a steady train of supply ships. We have been blindsided by the melancholy of the second generation, who felt alienated. They knew our planet only from movies and books. The third generation barely existed… We have scrambled to mount a mission to bring back any who wished to return.

After this debacle we have refocused on concrete problems. To avoid ecosystem collapse a decision was made to concentrate the population in several thousands of megalopoli. We have built upwards, but remained thethered to the ground. New materials we discovered enabled us to reach new heights, and build kilometers tall skyscrapers finally worthy of their name.

Our magalopoli were self sufficient, and the planet was able to recover.

The disater came in the form of a swarm of comets. Just water.

As the first group struck our planet, it turned into vapor, which quickly condensed. It took 50 years for the deluge to end.

By our calculations even the tallest peaks would not be safe. Our only hope was that the topmost floors would remain above water, so we have moved upwards and upwards. Trying to keep the power and internal farms on. Population has been reduced to almost nothing.

Today, pockets of civilisation survive scattered around the sunken megalopoli archipelagos. Thick clouds and ever present fog reduce our solar capacity to an accessory level, we have barely any fossil fuels left. Communication between

Then, one day… something has woken up.

Issues with space-age 4x games

I hope you have enjoyed this introduction to my 4X game project.

My ultimate goal is to fix what I deem to be the most glaring plot issues with space 4x games.

• Speed of light is almost always ignored.
• You as a leader are always omniscient and can order a ship lighyears away to do something immediately, there is no communication delay.
• Combat in games is mostly done on a 2D grid.

I’m setting it on an oceanic world to use sea vessels and islands instead of like sail ships and islands instead of spaceships and planets. The speed problem disappears and a 2D grid makes sense. Travel times can be justified by distance between islands.

• Scales make little sense, a planet in MOO2 can produce one building or spaceship over a span of several years, while having millions of inhabitants. (This was more realistic in MOO1).

The island populations will be relatively small, counting in hundreds maximum. The turn scale will be probably in weeks. So say, if you build a factory in 12 turns, that makes 3 months.

Speaking of “building”. Things in Thalassa are built by first collecting functional components from the sumberged buildings and then assembling them on a rig. This makes building stuff faster than one could imagine.

• Technology progresses one invention at a time (again, MOO1 did this better).
• Technology is not exploratory, player can always choose what they want, e.g.: give me a plasma cannon next.

In Thalassa you do not really invent new stuff, rather, you re-discover understanding of previously existing technologies, to be able to assemble them once again.

My goal with Thalassa is to make managing people the central point of the gameplay.

• Each island starts with a neutral population

• Population growth is slow

  • Population can increase either in the neutral camp or yours

• Your population increases by ‘recruitment’ not by birth

  • Recruitment can be volutary, forced or special (mind control)
  • Your population can be kept by different means (wellbeing, coercion, special…)

• Your followers can pop up on islands you don’t control

  • Same for the other cults
  • They can be overt or covert

• Genocide is frowned upon

• Morale - affects productivity and recruitment

• Loyalty - affects retention rate and spying

• Food, and health is global to island (archipelago)

  • You can provide or hoard, to the benefit/detriment of the neutral population

• Neutral population can help defense or offense, depending on your standing

• Neutral population can be more or less suggestible (depending on the living conditions)