Status of the Wiki

[Michael Wimble]

The Wiki still exists at https://wimblerobotics.github.io

At the bottom of every page, you’ll see:

I’d appreciate it if you’d always click on one of the buttons when you first see that page, or every time you see the page. What happens is that “Correct”, “Useful” and “Not useful” update a database. If you click on “Correct”, you’re just giving me feedback that you read the page and it all seems correct. This is great feedback that tells me I probably don’t need to worry about errors on the page. Remember that some (not all) of the content is AI-generated. I’ve looked at a lot of the pages and they seemed correct to me, but more eyes make it better. 

“Useful” and “Not useful” tell me whether this page should even exist.

The “Needs correction” page brings up a dialog where you can tell me what’s wrong with the page. One of two things happens: either your comments just get gathered into the database for someone to review and possibly act on, or you can actually create a Pull Request, which will get someone’s attention a whole lot faster. No personal info is gathered, we won’t know who you are. But, as will all things, if you can’t be constructive and polite, no one will act on what you say nor ever want to be your friend. You are supposed to be functioning adults, do the function thing in an adult manner.

Behind the scenes, there is a whole lot happening. I have created a process where I might scan the last three years or so of the e-mail repository for the club and auto-generate more Wiki content. I’m tweaking the process. A small number of people have looked at possible Wiki content generated by this process and provided me with some feedback. There is a maintenance web page which tells me stats from those buttons. That will let me know where content needs adjusting. Those who volunteer to help maintain the Wiki will get access to that page.

Remember that you can already contribute to the Wiki. The process is described on the home page.

So far, no one has volunteered to help with this project. Which is fine. In any volunteer organization, only 5 people seem to ever rise to the call for help, and those same 5 people do so for years at a time. Just look at the history of officers for the club. The only implication is that I have spent a fairly long time working on this so far, and my interest in maintaining it will vary. Those who bribe me will get responses and work on my part a whole lot faster than those who don’t. This has always been true even for those who I give private consulting to via Zoom.

But, and especially for those who haven’t yet had time to actually look at and try to solve a problem using the Wiki, and the Wiki is very specifically designed to help you solve problems, there is a fair amount of high-quality information in it. I don’t want to hear any of “Oh, it’s all AI slop”. It isn’t. The Wiki points you to landing pages that have detailed answers, and it even summarizes the key points before you go to the landing pages. In a short number of hops, you might find good information about how to build or fix your robot. I use this Wiki myself, and I don’t have a lot of time in my life to waste on low-quality web resources. Excuse me while I pat my back. Ahhhhh….

I don’t know if we’ll ever link this to the hbrobotics web page. It probably won’t happen unless people use the Wiki and request the club officers to add the link. So, until then, save the link above. Feel free to provide feedback. Those who bribe me and those who are willing to help maintain the Wiki will reap just rewards. Everyone else will be dealt with by Santa’s Elves who are constantly monitoring your worth. Only he can really know if you are bad or good.

And, ultimately, if no one finds the site useful, it will go away. <sniff> Meanwhile, the site is new and will continue to have significant changes to it before it settles down.

E-mail me feedback directly unless you think a handful or more of the 1000+ members on this mailing list will be interested.

[Lloyd Dickinson]

Thanks for all your hard work on this.

--
You received this message because you are subscribed to the Google Groups "HomeBrew Robotics Club" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hbrobotics+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/9332BB51-8D40-451A-8316-433408A22DDA%40gmail.com.

[Chris Albertson]

First off, the idea to use AI to turn emails into wiki pages was great.   It solves a chicken/egg problem where content attracts more content.  I never would have thought to do that.  But the problem with AI summaries in general is they read as if written by some guy named “Captain Obvious”.  Who writes things like

 

"If you look upward, you see the sky, it is mostly blue, but at night it gets dark, and BTW if you drop a rock, it will move downward.” 

It is 100% correct and the grammar is OK but you learn nothing by reading it.  The text does not solve any problem.

I think a better measure of goodness in the Wiki is “Did this page solve your problem” or “what did you learn from reading this page?”  Perhaps some more prompt engineering can help.  Ask the AI to create a problem-solving handbook that is organized hierarchically by technical area?  Tell it to start with each article with a simple summary for a general audience, but then drill down to sections targeted to expert-level readers, pulling in technical content from outside the emails  Can it even do that??  I don’t know.   Look at the better Wikipedia articles.  They follow this plan.

I am actually using AI-generated summaries to study two unrelated subjects. (1) computer networking, especially the new Thread 1.4 specification and routing IPv6 and also (2) to follow Yann LeCun’s latest work in AI.   The networking summaries are useful.  The AI just states facts and provides links to more details. But with #2 all I seem to get are summaries aimed at a 5-year-old or summaries of ignorant chatter scraped from social media.  I can fix this only by giving the AI very explicit instructions to summarize one paper for a specific reader.  But that requires me to first find the paper.    Point is the same AI can do good or poor work depending on the subject areas.

Maybe what I want is not a Wiki but an AI that can give specific advice.    For example, “ my motor speed is oscillating, what is wrong?” and then the AI guesses you are using PID, and the P value is very high. It writes a PID tuning article targeted at your technical level.

What I really want is a university librarian.   When I was in grad school, I could take a summary of my research requests to the library and talk with a real person.  She would go over the notes, mark them up, and tell me “come back in a few days,” and then she would have a collection of papers and articles I would never even have thought to ask for. 

On May 15, 2026, at 2:36 PM, Michael Wimble <mwi...@gmail.com> wrote:

The Wiki still exists at https://wimblerobotics.github.io

At the bottom of every page, you’ll see:

[Sergei Grichine]

Michael,

When looking at the Wiki (and discovering how it is generated from https://github.com/wimblerobotics/wimblerobotics.github.io) I, naturally, looked at the references to my "slgrobotics" repos.

This is represented by the following "agent prompt page" (or whatever it is called): https://github.com/wimblerobotics/wimblerobotics.github.io/blob/main/sources/slgrobotics.md

As a result, the five generated Wiki pages refer to my repos (thanks!):

• https://wimblerobotics.github.io/devops/jazzy-desktop-and-sbc-setup-for-robots
• https://wimblerobotics.github.io/devops/ros2-multi-machine-networking-and-discovery
• https://wimblerobotics.github.io/hardware/diff-drive-base-control-with-ros2-control-and-arduino
• https://wimblerobotics.github.io/hardware/rtk-base-stations-for-outdoor-robots
• https://wimblerobotics.github.io/navigation/gps-rtk-and-outdoor-navigation-in-ros2

They all contain links to my "robots bringup" root page, which is not optimal for navigation. I don't know how to point them to a better source—specifically, my Wiki: https://github.com/slgrobotics/articubot_one/wiki Does it make sense? 

Also, I'd like to suggest some other useful repos for the prompt page (at your editorial discretion, of course):

• https://github.com/slgrobotics/outdoors_loc_nav
• https://github.com/slgrobotics/ros_battery_monitoring
• https://github.com/slgrobotics/slg_bt_plugins
• https://github.com/slgrobotics/camera_publisher
• https://github.com/slgrobotics/face_gesture_sensor

I'd guess many Club members and visitors have repos they would like to share and have indexed in your Wiki. You may want to explain the easiest process for doing this. Is it a PR? A direct email like this? Where do I send your $20? ;-)

Overall, the more I look at your Wiki, the better it gets. Keep up the great work!

Best Regards,

-- Sergei

To view this discussion visit https://groups.google.com/d/msgid/hbrobotics/BE6B3FDB-668D-4D7E-A856-705AA9F4B329%40gmail.com.

[Michael Wimble]

I’ve made Sergei a collaborator on the project, without his permission, despite his probably howls of protest.

If you want your content included in the Wiki, you can:

• Post pull requests. No one is yet maintaining this besides me. My response will be sporadic. Note that Sergei knows the correct formula for getting a fast response, and I await his $20.

• Convince me that you have a lot of good content to add. For instance, Sergei has a huge trove of meaty information—if you haven’t looked at it, then it’s because you really haven’t tried very hard to find answers to your problems (well, his site only deals with robot-related problems). Then convince me that you are a good, upstanding citizen and that you play nice with others, understands git beyond the basics of being dangerous, and I may let you be a collaborator. Until I don’t. I’m an opinionated programmer and suffer fools in proportion to the amount of money sent, robot goodies sent, or cookie sent.

Otherwise, those who are actually part of our ROS 2 community for this club that turn up for at least the Tuesday and Thursday zooms know that my book, “ROS 2 for Mere Mortals,” is driven by questions that get posed during the Tuesday zoom. Most of the non-trivial questions point to deficiencies in the (barely useful) ROS 2 official documents. I take those questions and do the hard research, which usually involves just reading the actual code base. It’s not practical for everyone to do so if you aren’t an actual grey-haired programmer, so I do it for you. Then, I read the research papers that back the code. Then I try to write a chapter that addresses the gist of the problem and call out all the missing/hidden data from the code and explain, where needed, how the underlying algorithms work. Since my days of being a principal writer for Byte Magazines, back in the days when we switched from kerosene-powered computers to hobby computers powered by energetic electrons, my technical writing has always been an attempt to allow mortals to “get to the damn point” rather than forcing everyone to join Homer on an Odyssey in order to trick the Gods into giving up sacred information.

I’d love to get more content in the Wiki. I’m about to turn 76 in a few days and constantly have to decide where to expend my mortality. This is interesting to me for a small number of hours yet. Here’s your chance to be famous, to increase your change of scoring some Sadboys ice cream. So, contribute.