Request for Comments

[nicholascgilpin]

Are there any changes that should be made to the initial plan before I start coding? It's a startup style quickstart.

The initial plan is to import a bunch of data from cooking recipes (easy to market), build demo display inside a Notebook, and use that as a marketing tool to 10X the volunteer team again.

When there's no time for coding, we can:

• Post a link to the project description in related communities
• Add any resource-process data sources to the wiki

[nicholascgilpin]

@oliverevans96 @avianworld @luizeduu @amoffat

[amoffat]

I like the demo project and agree it will be necessary for marketing. Imo, the thing that will make or break this project will be the tooling that volunteers will use to contribute. If it's not incredibly simple for a layperson to contribute, they will bounce away and the marketing momentum will be lost.

I should clarify, these "volunteers" that I'm referring to are the ones who would build out the demo project further. In you original post, I think your "volunteers" refer to engineers that will help scaffold the engine?

[uwu64]

Hey, I'm more of an electronics engineer without any experience with developing software projects. I really can't contribute much to the code or planning (sorry!)

However I had ideas for recipes for semiconductor fabrication (specific to particular processes) which might be useful to demo some more "advanced" use case. (I would certainly need to do more searching around before I could write content for that)

Another idea.. How to make: Rolled film capacitor

Ingredients: Insulating film (or plastic or any material with sufficient dielectric quality) Conductive sheet (aluminum or any material that's inert and conductive)

Instructions:

1. Layer a piece of insulating film on top of conductive sheet
2. Layer another piece of conductive sheet on top of insulating film, making no direct contact and keep enough distance from first conductive sheet
3. Repeat as desired for increasing capacitance
4. Optional: roll and attach electrodes for more compact form factor

[OliverEvans96]

Have you considered using RDF / semantic triples for describing the relationships between nodes?

Also, are you familiar with the ontology editor Protege?

[OliverEvans96]

Also, keep in mind that most things can be created multiple ways.

[nicholascgilpin]

I like the demo project and agree it will be necessary for marketing. Imo, the thing that will make or break this project will be the tooling that volunteers will use to contribute. If it's not incredibly simple for a layperson to contribute, they will bounce away and the marketing momentum will be lost.

I should clarify, these "volunteers" that I'm referring to are the ones who would build out the demo project further. In you original post, I think your "volunteers" refer to engineers that will help scaffold the engine?

GUI's take a lot of man power, so my plan was to snowball the manpower:

1. 10 engineers chip away at about 10 hours of work (the Notebook)
2. The notebook helps us market to get 100 engineers
3. 100 engineers chip away at 100 hours of work (the GUI)
4. The GUI helps us market to get 1000 regular people.
5. 1000 regular people can sustain the project (1% rule social media) @amoffat

[OliverEvans96]

I propose a real-time chat about this for whoever's interested on Monday 2022-03-21 at 6pm US Pacific time: https://meet.jit.si/UniversalBuildTreeRFCDiscussion.

[OliverEvans96]

https://github.com/nicholascgilpin/universal-build-tree/blob/027353db3fad153893831223c1e38ca2ce3a8d73/README.md?plain=1#L44-L46

It might sometimes be hard to determine a single "official" name for a resource. What something is called depends both on the context and the speaker/reader.

• context: e.g. salt for cooking vs sodium chloride for chemistry
• speaker/reader
  • regional vernacular differences (e.g. pop vs. soda, elevator vs. lift)
  • translating to other languages

This reminds me of disambiguation on wikipedia, although this is sort of the opposite - a thing with multiple names vs. multiple things with the same name.

[OliverEvans96]

Also, resources/processes can be described at various levels of detail.

• How do you build a house? Well, what kind of house?
• How do you you cook a pie? Well, what kind of pie?

In a somewhat similar vein, many processes are hierarchical. Individual steps in a procedure can likely also be described as procedures.

[OliverEvans96]

https://github.com/nicholascgilpin/universal-build-tree/blob/027353db3fad153893831223c1e38ca2ce3a8d73/README.md?plain=1#L57

Not every process happens on the order of seconds. You gave a good example - the chemical reaction to produce sugar. On the other side of the scale, how long does a star take to form? That might cause integer overflow if measured in seconds :)

It probably makes sense to use a float w/ units for durations.

[OliverEvans96]

This might be a bit of a separate topic, but I'm also interested in deconstructing resources into other resources. I think this is valuable for recycling. I think your framework is sufficiently general to cover this as well. A "destructive" process also has a list of inputs and a list of outputs.

Although this would likely introduce cycles in the graph, so it would no longer be a DAG. Maybe acyclic is not a necessary restriction?

[uwu64]

Destructive processes depend on how good you are at destructing something right? You can rip out the guts of a PC to get power supply and memory and processor, and maybe desolder individual components like capacitors or crystals, but its kinda hard to get all the atoms back (would require a lot of energy I believe)

[OliverEvans96]

I agree. Apparently not every constructive process can be reversed. But there do exist useful destructive processes.

And to your point, the form and quantity of energy required may be a valuable parameter for many processes.

[nicholascgilpin]

Good comments.

To make thinking about the current design easier. I stole this dependency graph model from functions. It's a program only using functions, but in graph form. It might be officially called a "data flow graph", but I'm not sure.

Reponses:

• RDF / Semantic Triples - I'll look into these. The file format and tools look useful. Can dependency relationships be modeled in a knowledge graph? Dependency graphs seem to also used semantic triplets though.
• unambiguous official names - I believe this is covered by the unique ids?
• hierarchical - DAGs, aka Trees, are good for representing hierarchical nested details.
• time in seconds - definitely an issue. Any ideas to fix?
• deconstruction - Breadth first search in the reverse direction would show the deconstruction steps for a product.
• construction - steps can be shown via topological sort on the BFS deconstruction

[nicholascgilpin]

My schedule's pretty booked until June, so I'll be back then move things to the next stage.

Feel free to do investigations, add wiki notes, build prototypes, or fix the specification before then.