Jurisdiction
There is no Robotics Division.
There will also not be another meeting about creating one.
A robot is already an administrative pileup involving Circuits, Local AI, fabrication, software, mechanical systems, sensors, motors, vision, and at least one component ordered from a website whose product documentation consists entirely of a JPEG.
When a machine can sense, decide, move, manufacture something, and then contribute to the design of a better machine, nobody gets sole custody of it.
That mess is the Iteration Department.
Please stop submitting transfer requests.
A Brief History of the Iteration Department
Historians generally agree that the Iteration Department did not exist prior to its creation.
This was considered one of its strongest early accomplishments.
For most of technological history there was no need for a department devoted specifically to iteration.
People built things.
The things were bad.
They built less bad things.
Civilization occurred.
This arrangement worked reasonably well until several technologies began advancing at the same time and, more importantly, began becoming useful to one another.
Software started writing software.
Machines became cheap enough to manufacture useful parts on a desk.
Open hardware made more designs available to inspect, modify, and build.
Artificial intelligence began participating in coding, engineering, design, research, debugging, simulation, fabrication, and the mass production of very confident wrong answers.
The traditional boundaries between disciplines responded by becoming suggestions.
A programmer could design an enclosure.
A designer could build software.
An AI could write firmware.
A printer could manufacture parts for another printer.
A circuit could measure the machine producing the enclosure containing the circuit.
Nobody completed the required interdepartmental paperwork.
Somewhere in the resulting administrative failure, the Iteration Department appeared.
No surviving documentation identifies who authorized it.
This has made shutting it down surprisingly difficult.
Why Now?
There have always been tools that make better tools.
An imperfect machine helps build a more precise machine. That machine manufactures better components. Better components produce better computers. Better computers help design better machines.
Civilization has been exploiting that loop for a very long time.
What's changing is that the tools are beginning to participate in the loop themselves.
Software can inspect, test, modify, and improve software.
AI can write code, operate tools, design experiments, search enormous bodies of knowledge, evaluate results, and assist the research producing the next generation of AI.
Printers manufacture upgrades and replacement parts for printers.
Digital systems generate physical machines. Physical machines generate measurements. Measurements improve models. Models generate new designs.
None of these ideas is entirely new.
The interesting part is that the loops are starting to touch.
AI improves software.
Software improves AI.
AI designs circuits.
Better circuits make better AI possible.
Better fabrication produces better electronics.
Better electronics control better fabrication.
Each system increases the capability of the systems around it.
For most of history, humans were the connective tissue holding those loops together.
We still are.
Management would like that sentence emphasized.
But perhaps not quite as exclusively as we used to be.
For a brief and particularly weird moment in technological history, iteration itself appears to be acquiring tools.
The Department intends to observe the situation until somebody asks us to leave.
Organizational Outlook
The Iteration Department makes no claim that this organization will exist forever.
Statistically, websites disappear.
Projects get abandoned.
Domains expire.
Dependencies break.
Maintainers discover hiking.
There is every possibility that several years from now the Department will consist of four expired SSL certificates, a broken GitHub Action, and one search crawler repeatedly requesting /wp-admin.
But there is another possibility.
For a peculiar little window of technological history, powerful AI, desktop manufacturing, cheap electronics, open-source software, enormous public knowledge bases, and absurd amounts of consumer computing power are becoming accessible simultaneously.
A single person can build things that recently required a laboratory, a software team, a machine shop, or at minimum an adult.
That seems worth documenting.
Possibly important.
Definitely bad for available shelf space.
Operations will therefore continue until:
- the experiment ends,
- management discovers us, or
- somebody finally labels the drawer containing the USB cables.