@heise_ct

ich hab KI genauso wenig jemals mit ner Schippe in der AHnd gesehen, wie ein 50-€-Schein. von wegen, das Geld arbeitet für uns.

For mobile users: please decrease the sensibility to around 0.2.

For those interested on how I made the game:

I have always wanted to make an arcade game like Tetris, space invaders and pac man. You know, games that are simple but can be replayed forever?

But the idea of a game about keeping a pendulum upside down only came up when I saw a video of a robot doing it. It was a car and it moved to the left and right so that the pendulum never got too close to falling.

The hardest part was the ui. I'm very bad with ui for some reason
Maybe I'm just bad with css.

In order to make it a little more than just a bottle in front of a black screen I needed to add backgrounds, so I learned how to program a shader with the website shadertoy.

Making shaders is surprisingly easy and fun, although I struggled at first.

That's about it.

@ct_3003_team

sehr interessant, gerade bei den ganzen Thema "KI kann man nicht vertrauen, es halluziniert ja [ohne Toolanbindung] zu viel" ist eine Wissensdaten super.

Welche man dann verwendet ist dann natürlich eine hitzige Frage, aber generell überhaupt all sein Wissen in einer Quelle zu bündeln und einfach per lokaler KI dieses auszuwerten ist Klasse.

vermutlich wäre es dann besser viele Sachen per Texterkennung vorher zu digitalisieren.

@ct_3003_team

was ich mich aber gerade bei der MCP Anbindung Frage, wie verhindere ich Probleme durch die KI. Heutige LLM KI ist systembedingt kein zuverlässiges Programm.

Ohne Zugriff aufs Internet und ohne Kontakt nach Außen wäre die KI natürlich weniger Angriffsfläche ausgesetzt.

Aber sie könnte trotzdem sich mal verrennen oder die bedienende Person falsch verstehen, wie verhindert man das sie Sachen einfach so löscht, überschreibt, etc? Also in der Wissensdatenbank

I don’t know anything about the industry but I know the guy who made Stardew Valley was a one man show. Concerned Ape.

I believe it’s possible but is a steep hill to climb.

This got a bit narrative/logging what happens part way through, but hopefully it's still useful to you 😉 Really that's just a mark of me enjoying my experience with the game, so thanks for sharing it!

v. 0.1.264
1. The options menu feels cramped and overly narrow; visually it looks like the middle third of the previous middle‑third main menu.
2. I appreciate the instant tool‑tips for “Initiate Launch”, “Watch”, and “Fast”.
3. The pre‑launch phase reminds me of Oregon Trail—in a positive way.
4. The label “Research” under Ship Condition is confusing. It functions more like character traits or options during “character creation”. When I see “Research” in a game I normally expect long‑term goals that are pursued by spending resources.
5. When selecting research before departure, it would be very helpful to be able to “unselect” or toggle the chosen options so I can experiment with different builds and see total resource usage before committing.
6. Under Ship Condition → Systems, the order of the three entries should match the summary above. The summary lists “Food / Power / Hull”, but the Systems section lists them as “Hull / Power / Food”.
7. I liked the hidden‑room ship event that offered choices to force entry, hide it, or disclose it. I chose to disclose to everyone, but I was hoping for follow‑up events that branch from this decision rather than a single, isolated occurrence.
8. Bug: Year 35, Sensors offline, arrived at original destination. After surveying two of three worlds (ice world and volcanic; I skipped the gas giant), the UI kept showing the three survey options while text was still being audibly typed in the background. Selecting “Reject all and keep looking” then presented a separate reject‑menu for each world, forcing me to reject each one individually before I could proceed. A better approach would be a greyed‑out option at the bottom that says something like “You may only move on after rejecting each option individually” - and a choice of order to reject them in or something.
9. Bonus bug: After picking a new destination, the log reported the habitability of the “Gas Giant Moon” as 43 % even though I never probed it, so I should not have learned that value.
10. Confusion: In year 36, immediately after plotting a course to a new system, I received a “33‑deaths, Crisis‑level attrition – all life‑support systems reviewed” message with no directive or event. The status read Food 28 %, Power 35 %, Hull 42 % and all drifts ≤ 20 % except AI 46 %. Advancing another year gave another 32 deaths, still crisis‑level and still no event. Switching the ship to Cryo priority and advancing again resulted in 24 deaths. Hull decreased by 0.1 %/yr, Power by 0.6 %/yr, Food by 0.3 %/yr; population was 876 with an actuarial projection of a 2.5 % annual loss, which would be about 22 people per year, yet I was only losing 1 person per year. This inconsistency suggests a possible bug in the loss calculation. Earlier, the pre‑launch projection estimated about 600 survivors, which does not match the actual outcome.
11. I actually lost 36 people that year, not 22. After launching a probe, Power fell to 24 % and the actuarial projection changed to a 4 % loss per year, which aligns with the observed loss. This appears to be a direct result of the probe rather than a projection.
12. I lost 34 people during a “Voluntary Skills Census” event; there was no clear reason to avoid the full reallocation, so the outcome felt arbitrary.
13. Deaths have settled to roughly 18 people per year. Power is now 26 % and an event called “Power Rationing Debate” appears. I must choose between increasing AI (currently 47 %) or accepting a power reduction. I opted for the equal‑ration path (communist‑style), which had no negative side effects and reduced Power to 22 %.
14. Again I am confused: I lost 27 people, but Hull suddenly started regenerating at +2.9 %/yr and Power at +7.4 %/yr, while Food continued to decline at ‑0.3 %/yr. This coincided with an event titled “Shared Dream Phenomenon → Investigation”, which yielded no benefits except a reduction in AI. The following year Hull and Power returned to their prior slow decline rates.
15. Before departure, the initial habitability estimate was ≈27 % for the second system, but upon arrival, yet the system contained only three barren rocks with ≤ 3 % habitability. We moved on to the next system, (15 years with an estimated 20‑something % habitability) as obviously we cannot survive on a rock, so we must continue the journey.
16. Year 048: 27 deaths and a “Suspected Cause: Food Shortage” (Food at 23 %). I triggered the emergency intervention “Ration Proto”, but deaths remain around 20 people per year.
17. Year 051: 24 deaths, now the “Suspected Cause” is listed as “Life‑Support Stress”. Cryo remains active, Life‑Support system is nominal at 83 % Power, but Power is low at 13 % (mostly due to Cryo). The decline rate is ‑0.6 %/yr with an estimated 22 ‑year remaining.
18. Ship Condition reads “Critical; Cryo: Compromised” with no clear explanation. I formally recognized a spiritual practice, raising Ideology to 56 % for no reason except to see what happens.
19. Bug: “Notable Congregation Member Deceased – Very Quintero was a Scientist (Founder) who served on this vessel for 59 years. She was present at 0 recorded incidents during the voyage. She once said she just wanted to make sure the congregation survived the hard years. She was only there for 0.” Funny bug, but not reasonable numbers.
20. Arriving at the second system, the Ice World showed 18 % habitability. With Food 27 %, Power 6 %, Hull 46 % and a loss rate of ≈ 20 people per year (population ≈ 320), I landed. Surface conditions: Atmosphere 40 %, Water 40 %, Temperature 12 °, Resources 17, Biohazard 12. No accessible surface resources led to a construction crisis and extreme cold, prompting the Founding Directive “Shelter”. I also selected “Open Assembly”. The result is “Desperate Landing” with Pop 320, Diversity 95.1 %, Regression 8, Class Drift 0, Hull 42 %, Planets rejected 6. Key decisions (not mentioned above): Landing Protocol – AI‑Managed; Vessel Contact – Arrange Formal Meeting. (The next year we arrived at the final system and the other ship vanished from events, so I’m unsure how consequential that decision was.)
21. Description issue: the game claimed that ship‑born people outnumbered the original launch crew, which is false. The mission record shows 1,000 departed, 320 arrived, 21 ship‑born recorded, and 219 named deaths.

Just the suggestions from above, for your convenience:
* Expand the options menu width to avoid the cramped “middle‑third” appearance.
* Rename “Research” under Ship Condition to a term that means selected technologies, not long‑term progress.
* Allow toggling/un‑selecting research choices before departure so players can experiment with builds and see total resource costs.
* Align the order of entries in Ship Condition → Systems with the summary order (Food → Power → Hull).
* Expand more on branching follow‑up events after major decisions (e.g., the hidden‑room ship event). Only have one‑off outcomes for simple closed-ended scenarios.
* Fix the survey/reject UI so worlds can be rejected individually without repeated menus.
* Prevent undisclosed information (e.g., habitability of un‑probed worlds when sensors offline) from appearing in the log.
* Offer clearer feedback on attrition events: show why deaths occur, or always provide options to intervene. OR if those death rates are typical, fix the bug of attrition being too low at mission start.
* Clarify and correct event descriptions that contain nonsensical data (e.g., “Notable Congregation Member Deceased” stats).