July 2018 Core Group Meetings

Jack Harich

Administrator
Staff member
#1
Well, July is going to be an incredible month! Monteserrat travels up to Atlanta! Scott marches forward on seeing where AI fits in our long-term vision. Jack tries to upgrade the prototype to multi-user database. And pretty soon we all start claim-check experimentation cycles, as Montserrat's research begins. That's a LOT of action!

Due to some really big events in Mexico, like the most important election in 10,000 years, we will be meeting on Monday morning at 10AM EST, which is 7:00AM PST and 8:00AM Montse's time. Here's our modest agenda:

A. Montserrat
  1. Update on finding the perfect room in Atlanta.
  2. Update on all sorts of things related to moving to Atlanta and starting the internship with Thwink.org.
  3. Thoughts on starting the research, in particular the claim-check experimentation cycles.
  4. Anything else?

B. Scott
  1. Update on seeing where AI fits into our long-term vision.
  2. Possible review of the latest version of the Candle Core Functions Diagram.
  3. Anything else?

C. Jack
  1. Finally got the framework deployed, after 11 days of work. Now that it's deployed, I can see an expert could have done it in one hour. Amazing. What can we learn from this experience?
  2. Although the framework is running on a website, I'm having trouble with getting truthratings.com to point to truthrating.org, and for truthratings.org to run on https and not http. This problem seems to have revealed that the domain name register I'm using is using obsolete software and won't admit it. I will attempt to fix this problem somehow.
  3. Had trouble getting a tool called GIT to work for version control. Finally got it working.
  4. On Saturday I was finally able to start putting the prototype into the framework. That's good news! We are back to normal development, where there are no big roadblocks due to bugs and other problems caused by other people that are often very had to solve or cannot be solved, and I have to take a different approach.
  5. My focus right now is to get the Truth Ratings System working for online multi-user database functionality. Once that's done, we will have no software distractions when doing claim-checks. This will take an estimated 3 weeks.
  6. I spent a few hours studying the WikiTribune articles and their fact-check project. My impression is this is floundering around and trying to reinvent the wheel. The last Talk entry was in March. The page history shows the page edits have slowed down to nearly nothing. The goals come across as a quick wish-list based on no serious understanding of existing fact-check industry structure. And so on. But the biggest problem from our point of view is new fact-check articles seem to have stopped. I'm not seeing any in their daily email newsletter or on their Stories page. They are, however, including their article on Fact Checking Guidelines in their newsletters, even though it is not evolving much at all and also seems like it too is floundering around. Thus we may not be able to get much traction claim-checking their fact-check articles and publishing our claim-checks there. We will probably have to cast a broader net.
  7. Then again, maybe we can work closely with WikiTRibune and others to implement an innovative approach to fact-checking. That could begin by, after we've taken fact-check training, asking WikiTribune and other orgs if they would like to participate in our claim-check experiment.
 
Last edited:

Jack Harich

Administrator
Staff member
#2
The election results are in, and the best man won! It was Andres Manuel Lopez Obrador by a landslide!

Here's the agenda for our June 8th meeting on Sunday at 8:00PM EST.

A. Montserrat
  1. Report on finding the perfect room and the big move to Atlanta!
  2. Report on how the claim-check experimentation research is beginning to solidify.
  3. Anything else?

B. Scott
  1. Remarkable insights, cogitations, and musings on raising truth literacy!
  2. Anything else?

C. Jack
  1. Report on software development project. Going well, one week into the three week conversion of the prototype to use the new database framework. Worked very hard, accomplished my self-imposed goal of getting the countries, organizations, and users tables implemented. Nothing to show yet, since it's local development and there's no need at all to update the public website.
  2. Still have a database framework weakness. Errors on the server are not handled correctly. This causes the browser to lockup for a minute, until the connection times out. Will fix this but it's tricky.
  3. Here's something for us to thwink about. We will soon be writing and publishing claim-check articles. But if we publish elsewhere, such as WikiTribune, we can't control the CSS styles used to style the article. As a result our colored boxes for Claim, Rule, Fact, etc will not show. Nor will the colored lines. The best workaround I can thwink of is to use (FACT) and so on to indicate a fact node, and underlining to replace the colored underlines. Plus we can insert an image for the Claim Check box. Not sure how to handle the callout boxes, like the one for "The Fallacy of Cherry Picking." Long term, we may be able to talk some publishing websites into supporting a custom style sheet for our articles. To make this easier I can program an Export Bolded Article command to generate the html for hopefully copying and pasting into an article editor. We shall see.
  4. I must say it's really exciting to thwink about how close we are to getting the claim-check tool done and starting some high-quality research that, realistically, could make a tremendous difference. Wow!
 

Jack Harich

Administrator
Staff member
#3
Well, the political news is getting way too exciting. But let's turn our attention to the agenda for our June 15th meeting on Sunday at 8:00PM EST.

A. Group topics
  1. Discussion on meeting format, agenda construction, etc. How can we best make our meetings productive? It's so easy to fall into the unproductive meetings pattern, with lots of wasted time.
  2. Discussion on option to start a Meetup group on the Politician Truth Ratings project. The initial goal would be to attract a small number of high quality people who want to engage in "citizens democracy" activist work, by helping on claim-check experiments. As Montserrat said, "Less is more." We only need to find a few more high quality people to collaborate with. A problem is the past meetup group attracted lots of low quality people, in terms of their ability and interest as it related to our work. Fine individuals, but terrible analytical activists, which seems to be a rare mindset....
  3. Any other group topics?

B. Montserrat
  1. Report on literature review work, research work.
  2. The very latest on the perfect room and travel plans.
  3. Anything else?

C. Scott
  1. Suggested topics?

D. Jack
  1. Report on software development progress. Slow but moving forward. Have now implemented storing 6 out of 7 tables on the server. But encountered a bugs explosion when trying to do the Analysis page. Have determined the "root" cause is a very high use of asynchronous calls to the database. This causes the code to behave counter-intuitively and unpredictably, which leads to bugs. The solution is to redesign the code to take an Edit and Save approach, rather than auto-save and no edit mode. Implementing that solution is my next step. (Update: Implemented the solution, a whole class of bugs disappeared. Success!)
  2. The principle here is "If a poor outcome occurs repeatedly, then something is fundamentally wrong. Find it, fix it, test it, and resume." I've found this to be a powerful principle over the years.
  3. I'm personally trying to network and find a few high quality people. May have found one on Friday, a lawyer. We shall see. Will be talking to an old friend on Saturday who's very interested in our work and knows a few people that may be worth contacting. Again, we shall see. (Update: Struck gold. JC Dollar, my cousin, is a retired agency CEO. He loved the Politician Truth Ratings project and rattled off 8 influential people who might be interested, as I took notes. I told him the online multi-user version of the tool would would be ready in about a month. At that point, he is ready to spend some time with me to more fully understand it, develop a plan of action, and then start working with me and his list of people. But we may do better trying to interest the fact-check community first.)
  4. Had an idea today on the accuracy claim-check experiment. Start with just Jack and Montserrat. Get accuracy down to close to asymptote, then add another member, Scott or Martha. Get down to close to asymptote. Then add another member, etc. Advantages:
    • This greatly simplifies things by reducing the sources of variation.
    • Reduces frustration with others who are added, since so much discovery of problems, and fixing them, has already occurred.
    • Allows us to get going with a very small group. We are not blocked by having to find more experiment participants at first. Later that may become a problem, but at least it won't be a problem at first.
  5. Note the "actual" experiment may not start until we have a smoothly running tool with protocols, documentation, and a training program with certification. We shall see. There will be a lot of discovery here, since we are pioneers in a breakthrough area of social system research.
 
Last edited:

Jack Harich

Administrator
Staff member
#4
The Eagle has landed! Montserrat has flown to Atlanta and is living happily in the perfect college style shared home. Here's the beginning of the agenda for our July 22th meeting on Sunday at 8:00PM EST. Montserrat and I will fill this in this afternoon before the meeting.

A. Group topics
  1. First draft research plan for the rest of 2018. Our high level project plan is below.
Project Goals – Phase 1 – Claim-Checks Research (10 weeks)

- Learn R.
- Engage with the tell-the-truth community. (WikiTribune, others)
- Complete experiment testing accuracy of claim-checks.
- Complete experiment testing impact of CC vs FC vs nothing.
- Submit two papers describing the experiments.
The papers wrap up details would slide into the rest of the year.

Project Goals – Phase 2 – UN Research (5 weeks)
- Montserrat would like to learn the 3 main Thwink tools.
- Make a solid start on the UN goals research project.

Project Goals – Phase Ongoing – Masters application (1 week)
- Select a program
- Application prep
- Submit applications

Personal Goals
- Learn the principle to set a realistic task deadline and then as soon as possible get 20% ahead of schedule.
B. Montserrat
  1. Report on the trip and the "perfect" home.
  2. I need to set up a work structure for week days. We will start with working together at the Tower at 10:00 AM, then lunch break, then two or so more hours, then go home.

C. Scott
  1. Discussion of the new thread on Artificial Intelligence and Truth.
  2. More?

D. Jack
  1. Report on multi-user online software status. The first version of this is working. Got all tables working in the real database. Deployed the Truth Ratings System to TruthRatings.org. Fixed numerous bugs. Added the feature to check for circular references of reusable claims. Scott and Montserrat, you can now log in. It's truly wonderful to finally have a multi-user online version!
  2. Just yesterday I discovered the need for a large modification. When you are marking an article node to be a Fact or Reusable Claim, you actually don't know when it will be unless you are familiar with all the Facts and Reusable Claims in the database. Thus we possibly need a new mark representing a Fact or Reusable Claim. We also definitely need a combined list of Facts and Reusable Claims to search and select from, to replace the default Fact or Reusable Claim with the "real" one in the database. I will have to give this some serious design cogitation. Any suggestions?
  3. My hands will be full with further software development. That's so challenging and critical it will be my focus. I'm past (I hope) all the hard stages. Now the challenges are mostly designing and implementing user interface improvements. I have a short list of improvements to do. More will be discovered as we go along.
 
Last edited:

Jack Harich

Administrator
Staff member
#5
Well, Montserrat and Jack have finished their first week working together closely in person, and it went very well. Here's the agenda for our July 29 meeting on Sunday at 8:00 EST. As usual, we can modify it before and during the meeting.

A. Project reports
  1. Montserrat's collection of research projects. During the week Montserrat and Jack worked on a goal and task outline of the projects, plus risk management. Montserrat will be giving a weekly projects status report. This meeting will be the first such report (with discussion) and will probably take up most of the meeting.
  2. Scott's ongoing artificial intelligence learning project. This is designed to glean insights into AI that can benefit the Candle and Truth Rating System visions.
  3. Jack's software development project. I only worked on this a few days, since the rest of my time went to planning and working with Montserrat on her mega project. In general, the development of the Truth Ratings System is going very well. I'm completely over the very hard part of learning about how to implement a real database. Now the focus is a rock solid tool for doing claim-checks. Currently I'm adding the new feature of you can search for facts and reusable claims in one list, select one, and replace the selected fact or reusable claim in the article you are editing. The project is a lot of tedious work that needs to be very high quality, so I'm in a high state of flow most of the time. If I'm not, then I do something else and freshen up!
Next come team member items:

B. Monteserrat
  1. Any topics here?

C. Scott
  1. Any topics here?

D. Jack
  1. I like to thwink ahead about what's next and spot the hard or high risk tasks. We've got a big one ahead. When we start writing claim-check articles, we will discover that for rules, fallacies are well researched, but non-fallacies are not. There is no good collection (?) of non-fallacies that we can use to serve as a starting point for our list of rules. Maybe we will find some, but it may be that we are forced to create out own collection, with names like "preponderance of evidence" and "statistically valid correlation with the evidence." This also gets into deductive versus inductive reasoning. Scott, can you mail those high quality argument books back to me? Thanks.
  2. I would like to personally mention that it's been fantastic working with Montserrat. She's a fast learner, a real perfectionist but not overly so, has high self-awareness, does great work, and sees very deeply into what the Thwink paradigm is all about and its potential impact. Welcome to the Thwinker's Club, Montserrat! Your work is going to make THE difference in getting the Thwink paradigm explored, improved, and ultimately, applied.
 

Jack Harich

Administrator
Staff member
#6
Meeting results: Excellent discussion on the agenda items, especially the first one. Montserrat has provided these notes on her topics:
The topics I covered yesterday on section B of the agenda were:

1. I talked about my thoughts on Scott's AI thread, mainly in the ways I think Thwink could collaborate in the future with nascent news projects, such as WikiTribune and Knowhere News.

2. I talked about the importance of emphasizing on the fact that Thwink's basis for objectivity is what the Common Good is, and that our definition for Common Good is an objective, "scientific" one. This needs to be perfectly understood, as it's the core of the idea of being able to assess what the truth is. I mentioned two examples of how a pro-immigrant narrative could seem to be a lie, and a narrative against immigrants could seem to be truth, so from a superficial point of view, a high CL for the 1st statement, and a low CL for the 2nd one could seem to be biased. It is not, and that's what should be very clear.