Establish a prioritization technique about raven HOT 8 OPEN

I've previously been a fan of MoSCoW

from raven.

This is great stuff. I haven't yet formed a preference, but here's how I think of it:

• Prioritization is in layers, with some things (like the nexus) at the very core of decision-making and functionality
• The innermost layer is a necessary prerequisite technology/decision
• The next layer is everything that is immediately dependent upon the core layer (blocked)
• So on and so forth up to the outermost layers

As RAVEN/MARAGI is a "cognition first" model, the output layer is actually one of the least important (still important) but it comes last due to the fact that all the cognition, planning, and decision making has to happen first. Another way to think of this is more like the OSI model with layers of abstraction.

Maybe this is something I need to do to help communicate the architecture. This won't dictate prioritization but it will help frame stuff... Let me take a crack at it.

from raven.

The architecture is a hot topic I'm keen to discuss more, as I can help put together different viewpoints for different audiences, and express the architecture through various frameworks (archimate, TOGAF, etc) and patterns. Sounds like some decision trees or matrixes could be helpful also.

To be continued...

from raven.

@timfoote, excellent work.

from raven.

Hey everyone! I had no idea these techniques and methods for prioritization of issues even existed. Defining a system makes sense knowing how OSS issues tend to balloon out of control as projects get traction .

My question is: Assuming the chosen technique or method can be well specified and deterministic. Can we prompt or tune a LLM like GPT to classify and label issues based on the chosen method? If so then github workflows can be used to automate initial issue labeling.

If it's possible and out of scope for this project. I assume there are a lot of projects that would find value in automating some of the chores around issue labeling and prioritization. Would it be a good standalone AI project?

from raven.

I'm happy with any of the above, but if I had to pick one I'd go for a Value vs. Complexity Matrix as they inherently lead to quick wins. That said, if used in isolation, some important features may be deprioritised.
Would a Value vs. Complexity Matrix with tags for MoSCow statuses be helpful?

from raven.

Love all of the prioritization techniques here! Not much to add here besides the fact I’m a big fan of MSCW and feel it’s a good all-rounder for product development.

Also, it may be beneficial to use Prioritization Quadrant in conjunction with MSCW for the initial development of RAVEN R1/R2.

from raven.

I really like the idea of combining mscw and the Complexity X Value matrix. This would allow us to both understand what we need to build now, and figure out who should work on what.

For example, 2 equally important items may be must haves, but one may be significantly lower complexity, and could be given to someone with less time, or less experience. It aligns well with the good_firsr_issue tag, which is a really valuable tag to have in open source projects.

If we didn't want the overhead of both, we could also just stick with MSCW, but enforce a tagging system of every item, 1-5, defining complexity.

from raven.