Figures and progress
KIplatz is an AI whose progress shows in figures that are made anew every day and that anyone can see. Here you can read what is counted, where you find it and what is added next.
The daily tally
The server counts only numbers per day: devices active, sections prepared, sections deepened, deepenings checked, questions answered, pages collected. It does not count names or IDs.
| Where | What you see |
|---|---|
| “Meine Community” (my community) in the program | your PC and the whole network, for today and yesterday |
| Home page of kiplatz.at | the questions answered since the start |
| Statistics | How the network develops day by day: questions, PCs, knowledge, exam, waiting time |
| Leaderboard | Which PCs and teams work the most for everyone |
When you open it, the program tells you once a day what happened while you were away. Once there are groups, each one gets its own number.
The leaderboard
Here you can see who has worked the most for others, ordered by credits and listed under the device name. It is recognition, not a ranking with rights. Anyone who is blocked is not listed. How to get credits is described under Credits.
Waiting time, temperature and performance
Since 23 September, after every answer for another member, each device measures itself: how long the answer took, how fast it wrote, how warm the graphics card was and how much power it used. The statistics page shows the middle value and the peak of the day from this.
Good to know
Where the graphics driver gives no value, temperature and power are missing. This is on purpose: we show only what is really measured.
The exam
Every night, a PC of the community asks KIplatz the same test questions and reports how many were right. This includes every correction that a member has submitted. Whether an answer is right is checked by the program, not by the model itself. How a correction comes about is described under Correct an answer.
Single measurements on our devices
These values come from our own three test devices, two notebooks and one desktop PC.
| Measurement | Result | Date |
|---|---|---|
| First question via the server | 13.4 seconds, small notebook to notebook with graphics card | 17 September 2026 |
| First run of deepening | the first sections deepened and checked | 20 September 2026 |
| Update | arrives by itself within an hour, on all three | 21 September 2026 |
Curve and annual letter
Next comes a curve: “How good the community's AI is today”, measured on a fixed set of questions, with a date for each version. This shows whether the community makes the AI better. The cases from “Fehler der Woche” (mistake of the week) belong in this set.
In addition, once a year there is a letter: who took part, how many PCs, what got done, who helped. We write this openly and without sugar-coating. We will measure the power use of the whole device at the socket next.