Typescript 2023-12-03

09:02:01

09:02:01 From read.ai meeting notes Gene added read.ai meeting notes to the meeting. Read provides AI generated meeting summaries to make meetings more effective and efficient. View our Privacy Policy at https://www.read.ai/pp Type "read stop" to disable, or "opt out" to delete meeting data.

09:03:57 From read.ai meeting notes Read has left the meeting. To request that Read rejoin the meeting, visit https://api.read.ai/sessions/01HGR9S697A172SR9JQMQ9ZB6A/rejoin

09:04:52 From Jeff Miller I have been out and around at Recurse Center and have been missing notes. Paul R. is really good with links! so the chat is shared commentary plus me trying to summarize points folks are making

09:05:58 From Paul Rodwell I’ll try and remember to upload the notes from the last few meetings

09:06:06 From Jeff Miller ty Paul! "You know my methods, Watson." boost and share!

09:07:19 From Jeff Miller "people in your server are talking about..." (Mastodon, from Ward's commentary, about being a timeline that repeats itself a lot) TwiX "For You" is getting better.

09:08:25 From Jeff Miller (for me anyway) ads show up in the fourth or fifth tweet in a screen that tactic actually works, blocking ads on Twitter "coupon clipping" labor in Twitter by blocking ads

09:09:43 From Jeff Miller Mastodon: following hashtags has been great! though #TIL is hits and misses (ongoing discussion of social media experiences)

09:10:58 From Jeff Miller Ward relates a 10-slide explanation from Tim O'Reilly on how to use Twitter -- it worked well for a while.

09:12:17 From Brian Yeah, can't be EXACT value, slightly less or slightly more.

09:12:20 From Jeff Miller Massimoso on exchange

09:13:31 From jan dittrich (er/he|they) https://en.wikipedia.org/wiki/The_Gift_(essay)

09:13:52 From Jeff Miller ty!

09:16:11 From Brian If you have power, you have an avenue for income...

09:16:18 From Jeff Miller (anecdotes about gifts and tax consequences)

09:16:22 From jan dittrich (er/he|they) aside: I love the PDP’s design

09:19:02

09:19:02 From Jeff Miller http://recurse.pixiereport.com/view/welcome-visitors/view/project-list/view/memory-card-creator

09:22:17

09:22:17 From jan dittrich (er/he|they) Jeff talking about: https://en.wikipedia.org/wiki/I_Was_a_Teenage_Exocolonist and weatherpixies

09:23:36 From jan dittrich (er/he|they) and https://gtoolkit.com/start/

09:25:20 From jan dittrich (er/he|they) "GT is implemented in Pharo. If you are not familiar with Pharo, take some time to follow A gentle introduction to Pharo Smalltalk."

09:32:17

09:32:17 From Marc Pierson I would like to introduce Gene Bellinger soon.

09:37:24

09:37:24 From Marc Pierson Perfect timing. Now Eric is here.

09:37:26 From Jeff Miller Pharo / Glamorous Toolkit / Lepiter Notebook / Moldable Development.

09:38:45 From Jeff Miller "a moldable toolkit for the user to develop their own tools for the problem(s) at hand, drawing from a rich library of partial tools, views, associations and collections" (reflecting on GT)

09:39:16 From Marc Pierson It has seemed to me that there is a back and forth possible between FedWiki and Wikipedia.

09:39:37 From Jeff Miller I have really been appreciating the JS console warnings for things like "oh, if you want to do that, you can set a property to make it more efficient".

09:39:42 From Brian To me, the FedWiki to Wikipedia gap is very large.

09:40:53 From Marc Pierson Kerry and I asked Gene to come here and show us what AI can do with CLDs.

09:41:08 From Jeff Miller "the conversation in the GT community isn't so much in shared channels and words, but in snippets of GT code, because their shared language is in mastery of operating the Glamorous Toolkit platform"

09:42:18 From Jeff Miller Marc introduces Gene Bellinger to the Wiki Sunday Explorers group. "The power user for Kumo, a persistent voice in the systems community about diagramming" "almost 50 years as a systems thinker until I wised up and became a storyteller"

09:43:37 From Eric Dobbs Right off the bat, I love the phrase “relationship models”

09:43:48 From Jeff Miller "I figured out a way of looking at a model and asking the right questions of them. I create relationship models. But don't they relate to everything? They do!"

09:44:59 From Jeff Miller "I create a model that describes what I think I know, and then elaborate it by conversing with people who know more details about the subject. I found that I could ask Bard and ChatGPT about subjects related to my subject of interest, which shortens the loop of refining the model." "Here's a demonstration of some software which I was introduced to recently, that I found interesting and compelling?" "AI Generated Causal Loop Diagrams"

09:46:30 From Jeff Miller https://causal-loop-demo.streamlit.app "It labels the connections with a note on the time scale for the operation of the relationship." "It recognizes some common loops, such as Success to the Successful"

09:47:56 From Jeff Miller Sample Causal Loop Diagram: "What are the future implications of capitalism in terms of wealth inequality and social cohesion?" "The people who developed the software are looking for use patterns and applications that fit it."

09:50:54

09:50:54 From Jeff Miller (Gene Bellinger narrates a demonstration of the Causal Loop Demo app, including how the diagram can be generated with relationship to a knowledge model encoded by the language model in GPT-4; that the models can be explored, tuned, and refined)

09:51:00 From Brian Same question I have for all ML generated models is how to verify the validity of the graph. If it's used to bootstrap a research area, that seems fine. If someone uses it to make a consequencial decision, that isn't very good.

09:51:24 From Jeff Miller time scales for "business tactics" too short for "making sense of the tool" ?

09:51:49 From Brian LLMs basically "average" and look for similarities.... so if it was trained on inaccurate data, it will reflect that in the results.

09:52:07 From Jeff Miller Gene says: "I'm fascinated by what the tool does and [how I can explore and evolve models.]"

09:53:14 From Jeff Miller Gene shows a list of numerous experimental relationship diagrams created during exploring the tool.

09:53:15 From Brian You mentioned something about the CLD to story part, so how do you take the generated CLD and turn it into a story.

09:54:23 From Paul Rodwell typical LLM, lacks any source references

09:54:52 From Jeff Miller (Ward reflects on experience talking with different causal loop modelers and seeing their approaches to building CLDs. "It's not surprising that the LLM is a body of content which expresses relationships among concepts and activities.")

09:54:56 From Brian Have you explored an area you are an expert in.

09:55:17 From Jeff Miller +1 on Brian's note: "is it good at something you know well?" (discussion about whether the results are plausible, and how well they're grounded)

09:56:04 From Marc Pierson This shows us how our crazy uncle thinks. Now we can better prune his narrative and keep what we want.

09:56:33 From Brian If it's used similarily to a psychic, then that is valid. But it can't be used as evidence.

09:57:09 From Marc Pierson Jay Forrester said that people are expert in noting relationships, but very bad at getting the direction correct.

09:57:09 From Jeff Miller Gene says: "about 1 in 10 models, it would get a relationship wrong in the sense, a + or -, but when I read the details, the description of the relationship was relevant, the program had misread the sense".

09:57:35 From Brian We shouldn't antropormophize chatGPT.

09:57:48 From Jeff Miller https://en.wikipedia.org/wiki/Jay_Wright_Forrester

09:58:00 From Brian Or it's inverse.

09:58:06 From Jeff Miller "causal direction, or sense of improving or limiting?"

09:59:06 From Brian Future steps for the graph are "abstractions" i..e. how to prune the graph for a given purpose.

09:59:34 From Jeff Miller Marc notes that it's common for people to choose models that have too short a timescale, or are missing key feedback loops. (re Jay Forrester's thoughts) Jan D. notes that LLMs have been helpful for brainstorming and considering alternatives. Marc says: LLMs are like my brilliant crazy uncle.

10:00:12 From Brian Depends on the balance of crazy and brilliant...

10:00:53 From Jeff Miller "I love to listen to my brilliant crazy uncle, but I don't trust what he says"

10:01:09 From Brian Might be text length...

10:01:33 From Jeff Miller Marc and Ward ask: "Where does the CLD app get data from the world? It seems to have some information that might not be in the basic LLM."

10:01:37 From Brian To me, this looks like it's reading the LLM response and pulling out key words, much like Jeff's index code, and turns it into a graph.

10:01:56 From Jeff Miller Ward's index code! (I just press the buttons, so far) https://kumu.io - the Kumu app

10:03:17 From Brian That graph needs some subsystems extracted out of it...

10:04:13 From Jeff Miller Gene B. observes: "here is what the LLM platform doesn't do -- it doesn't do a good job of telling a story about the pieces. For example, Unwanted Pregnancy Options, there are two balancing loops and another loop, but a connection that puts you in a vicious reinforcing loop that generates a shifting-the-burden structure."

10:04:19 From Brian Can you reproduce this work with your chatGPT program?

10:04:52 From Jeff Miller "To put these relationships in the original diagram would have made it more of a mess." "The LLM prompted and generated CLD gives me a draft to start from, which gives me something to ask questions of and let me explore a story that helps bring out the most interesting and relative parts."

10:06:04 From Jeff Miller "It only recognizes one- and two-loop archetypes of causal loops"

10:06:06 From Brian Cycle

10:06:18 From Jeff Miller "success to the successful" a name for a reinforcing loop of a particular type. archetype ~ pattern ?

10:07:37 From Jeff Miller "the archetypes are built from the evolution of the balancing and the reinforcing loops, for example, growth slows over time, growth here leads to decline elsewhere, growth limited by underinvestment, underlying causes not addressed, addiction pattern"

10:07:58 From Gene Bellinger https://kumu.io/stw/the-phoenix-project#frequently-recurring-structures

10:08:03 From Jeff Miller Once you recognize an archetype, you have strategies to deal with it. - Gene. Ward asks Marc: "Does Kerry uses archetypes" Marc: "I think so, but the clients don't find the language useful."

10:08:35 From Paul Rodwell Archetype => https://www.etymonline.com/word/Archetype

10:08:50 From Jeff Miller hmm "archetypes in causal loop diagrams" specifically? an example (PDF) of systems dynamics archetypes https://thesystemsthinker.com/wp-content/uploads/2016/03/Systems-Archetypes-Basics-WB002E.pdf

10:10:00 From Paul Rodwell from Jung - pervasive idea or image from the collective unconscious

10:10:06 From Gene Bellinger And? -> https://bra.in/7pDL6r

10:10:10 From Jeff Miller "Fixes that Fail", "Shifting the Burden", "Limits to Success", "Drifting Goals", "Growth and Underinvestment" so "archetype" here is more like "pattern" in our usage a recognizable configuration of forces

10:12:30 From Jeff Miller Ward: "What would we put in a two-day class about Smalltalk?" / "Here's this thing by Christopher Alexander describing how things connect" - Kent Beck Marc says:
Kerry would build a large body of documents
create a model from what she learned reading the documents
suggest where to ask questions to prompt interesting stories

10:14:39 From Jeff Miller "Everyone has to create their own model, otherwise it's a waste of time" - Marc, on Kerry's practice of prompting her consulting clients to build the model, by highlighting what she thought was most interesting in their system based on her learnings.

10:14:43 From Brian Until you get to the peer-to-peer rubber meets road level of work, the conversation is much more effective at the "The Heilmeier Catechism" level questions. https://www.darpa.mil/work-with-us/heilmeier-catechism

10:14:56 From Paul Rodwell Replying to "so "archetype" here ..." "model, first form, original pattern from which copies are made," 1540s [Barnhart]

10:16:22 From Jeff Miller Ward reflects on working at Nike in sustainability, and how taking the information present in their records of sustainability practice, and organizing it visually and graphically, with snippets of text and understanding, was seen as valuable -- and also was hard work, organizing for understanding. Replying to "so "archetype" here ..." Platonic! :)

10:17:43 From Brian We don't have AI, non of our current models are intelligent.

10:18:43 From Jeff Miller Eric reflects, unstructured, on his sense of the use of LLMs and CLD;

first, what's bothersome about LLMs as a thing, right now, in the current setting. Eric's work in incident analysis in software -- large, distributed systems with different teams responsible for different subsystems; we don't understand that the problems are significantly in the interaction on the edges between teams.

10:19:59 From Jeff Miller Eric says: the team I'm on does work across other teams, responsible for the various subsystems, to get a picture of where the expertise is present for understanding the system, and how it does or does not get brought to bear usefully on the incident. We'll ingest the conversations about the incident, who was talking with whom, and how did the insight unlock the solution -- or what false paths were followed, why, and how long? Why was a false path persistent?

10:21:36 From Jeff Miller Eric says: the LLM analysis is good at summarizing the sequence of events;

but it is bad at explaining why; the critical points are completely missed.

10:21:37 From Marc Pierson It is a description rather than an explanation.

10:21:46 From Jeff Miller (that's a Jerry Weinberg story!) (a group of people debugging some software, and the visible activity was largely irrelevant to the insight and solution)

10:22:52 From jan dittrich (er/he|they) @Jeff Miller: is there a link to the story? Sounds interesting!

10:22:55 From Jeff Miller "the metaphor of cost, to explain the latency of a query" @jan dittrich (er/he|they) I think it's in "The Psychology of Computer Programming" hmm

10:23:27 From jan dittrich (er/he|they) Reacted to "@jan dittrich (er/he..." with 👍🏻

10:23:54 From Marc Pierson Cannot see the difference between metaphor and the experience itself.

10:24:00 From Jeff Miller Eric describes a misleading metaphor, where it's possible to misread words used in a different sense -- that LLMs have exactly this problem.

10:24:17 From Brian LLMs lose a lot of the context and may select the wrong context, and then associate words incorrectly for that context.

10:24:29 From Marc Pierson I know lots of people that are no smarter than LLMs.

10:24:31 From Jeff Miller "The humans will be so busy and misdirected, that the LLMs will follow the confusion"

10:25:55 From Marc Pierson The CLD is an insight. The archetypes are further insights.

10:25:55 From Jeff Miller Eric reflects on Gene's model for unwanted childbirth, where Gene added relationships to the model that provided the key insight -- which the automated versions won't be able to do. "What are the forces at work in unwanted pregnancy", also a reframing around problems

10:27:22 From Jeff Miller Gene says: "Consider Ackoff's reflection -- wherever you discover symptoms, it's likely that the problem is elsewhere. The symptom is not the problem."

10:27:29 From Marc Pierson The drunk looking for his keys under the lamp light rather than where he dropped them.

10:27:37 From jan dittrich (er/he|they) Reacted to "The drunk looking fo..." with 😃

10:28:34 From Jeff Miller Eric continues reflections about LLMs - e.g. pointillism, "this is just a pile of dots, what turns this into a picture of a tree?" ("who is the Master who makes the grass green?") Ward says: "Is that a tree? I thought it was a mushroom cloud!" Eric says: "or is this a color-blindness test?"

10:30:07 From Brian It's the relationship of the dots into a structure that makes a pattern you associate with labels that are accociated with concepts or constructions. With those dots, you could reorganize them into 1000's of different pictures.

10:30:31 From Gene Bellinger As Bateson said... It's the pattern that connects.

10:30:46 From Brian So the dot's themselves, are not sufficient for the picture, it's the relationship or structure that is the important part.

10:30:56 From jan dittrich (er/he|they) Communication/Transmission: https://lchc.ucsd.edu/MCA/Paper/ingold/ingold1.htm

10:31:18 From Gene Bellinger We're very fluent in talking about what we think we understand.

10:31:35 From Jeff Miller Eric considers: "There are pages in the FedWiki riffing on George Lakoff's warning about communication as a conduit metaphor, from my brain to yours; but we're so fluent at looking past our misunderstandings to synthesize imaginary understandings from statistically plausible utterances." Marc says: "campfire conversations and cave drawings" - the conversation can be ambiguous; the drawing can capture a reflection which can be revisited and corrected.

10:32:31 From jan dittrich (er/he|they) Replying to "The drunk looking fo..." which is from "The Situation Is Hopeless But Not Serious" I think

10:33:10 From Jeff Miller Eric says: Like Kerry's approach, the interesting activity is the conversation in building a model together, not about bringing a model and then talking about it."

10:33:34 From Brian When there is an organization of dots that has previous value, then we can compare that organization with new examples and ask questions about how it is similar and how it is different and gain insight from that.

10:33:55 From Jeff Miller Ward reflects on pairing with Kent Beck on code -- it always felt like there were three of us in the conversation. "place, as important as any idea" - a notion of situated insights - from Ward.

10:35:06 From Ward Cunningham Email to ward@c2.com

10:35:08 From Jeff Miller Gene invites people to send email addresses via chat -- probably direct message is best for Zoom -- because of interest.

10:35:50 From Eric Dobbs Replying to "Gene invites people ..." eric@dobbse.net

10:35:56 From Jeff Miller In case other people would like to reflect on the LLM prompted diagrams, or where there might be potential. Marc: "They're easy questions, it took me about five minutes." Thompson Morrison's work -- providing for the needs of noncompliant learners in the classroom -- he thought of them as having more sophisticated needs than the compliant learners. (from Ward)

10:37:31 From Jeff Miller Thompson Morrison's approach, "wiki haiku", for creating modular pages which can be moved around helpfully in Federated Wiki as a knowledge cloud or knowledge base; Thompson's interest in explaining his approach, so that others could use it; he experimented with ChatGPT and asked it to describe the relationship among a couple of thing he was interested in.

10:38:33 From Marc Pierson Gene, do you use this query method? How is this related to that?

10:38:54 From Jeff Miller "I see that paragraph 3 and paragraph 6 in ChatGPT describe pages that I already wrote about this relationship!"

10:39:31 From Gene Bellinger Mark, the question is seldom simply like Capitalism though more like... what are the future implications of capitalism in terms of wealth inequality and social cohesion?

10:39:38 From Jeff Miller Ward's description of ChatGPT as a way to fill in the gaps in a plausible way, and then revising and refining it; "It feels very fast" - Thompson on ChatGPT.

10:40:07 From Marc Pierson “future implication”

10:40:44 From Paul Rodwell possible futures

10:40:45 From Jeff Miller (time scales of feedback and decision making in the presence of financial markets, property, and law)

10:40:53 From Marc Pierson I would like to learn what are the “best” ways of asking this new CLD LLM tool questions.

10:41:40 From Brian My mental model for chatGPT is two parts.
* Common encoding for words and images
* clustering method, like K-nearest neighbors.

10:41:41 From Jeff Miller Ward underlines the interesting proposition of having Thompson and Gene compare and demonstrate their approaches to LLMs as an assisting force for creative knowlege work. Thompson Morrison He shows up occasionally, gives us lots of ideas and energy, and shakes us up.

10:45:13

10:45:13 From Brian Don't remember if this book was from previous conversation or not, but just got http://cup.columbia.edu/book/storythinking/9780231206938 and will read it soon.

10:45:40 From Jeff Miller (mentions of Elinor Ostrom)

10:47:30 From Marc Pierson Game theory, non competitive games. A very big deal.

10:48:17 From Paul Rodwell Bruce Schneier - https://ash.harvard.edu/ten-ways-ai-will-change-democracy

10:49:03 From Gene Bellinger FYI: I have to leave at 2pm

10:49:08 From Brian One of the universities extended the prisioners delima in a number of itneresting ways...I'm not finding the study right now. A simplified version is that you should be vulnerable 1 time, and 2 times is very similar, but definately not 3 or more (in the contrived context of prisioners delima world)

10:49:51 From Jeff Miller Eric says: "I heard something new in Woods' appreciation of Ostrom's work -- the reason she got the Nobel Prize in Economics is that she proved that the Tragedy of the Commons and the Prisoner's Dilemma have the math wrong; that reciprocity is the rule, because otherwise we'd expect ruin to be frequent; biological systems and human systems have this resilience from reciprocity, where human-built engineering is fragile." "residuality theory", Barry O'Reilly

10:50:19 From Paul Rodwell https://www.blacktulip.se/

10:50:41 From Brian "risk" is another word that is very context relevant for understanding it. I think we are fortunate to have juries...

10:51:33 From Eric Dobbs Barry O’Reilly, Residuality Theory & my notes from a conference talk on the subject http://wander.dbbs.co/residuality-theory.html

10:51:46 From Brian That there is humanity in judicial system is important. If it were all logic, I don't think that would be a better place, even if it was more "fair" by some metric.

10:52:41 From Jeff Miller (mention of AI risks in law) "not proven" as a distinct verdict in Scotland.

10:54:06 From Jeff Miller Brian asks if it's equivalent to "not guilty"? (or a mistrial?) -> yields a retrial, possibly.

10:55:05 From Marc Pierson https://www.linkedin.com/in/chessscholar/

10:56:19 From Ward Cunningham Linkedin won’t show me that link.

10:56:59 From Paul Rodwell try https://uk.linkedin.com/in/chessscholar

10:57:10 From Jeff Miller John Foley, a colleague of Kerry's, who now teaches kids strategy in a private high school; a chess grand master at Cambridge. 

John Foley
 3rd degree connection
Strategic thinking, games-based learning, education and training.

10:57:20 From Eric Dobbs And here’s a thing from his profile. John Foley: http://www.chessinschools.co.uk/

10:57:23 From Jeff Miller from LinkedIn, seemed to work for me after a sign-in.

10:57:39 From Paul Rodwell https://chessplus.net/

10:58:04 From Jeff Miller Marc says: John Foley immerses the students in the kinds of situations which prompt strateging thinking. which prompt strategic thinking.

10:59:42 From Jeff Miller Brian reflects on the differences between playing games to win (hyperstrategic/tactical) and playing games to enjoy time together.

Marc says: John Foley tries to create non-competitive game scenarios for the students.

11:00:24 From Marc Pierson Cooperative games. Win-Win

11:01:28 From Marc Pierson Team competitions is a hybrid.

11:03:37 From Jeff Miller Ward reflects on the genesis of the PLoP conferences, with periodic "game time" interruptions from the game expert, George; it kept us in the mindset of "we're here to explore this new space together.' Ward describes one of these games, where you have a tennis ball, you have to toss it to someone, _first_ calling their name.

11:04:46 From Jeff Miller Once you've touched the tennis ball, you don't touch it again, so everyone has their name called out, eventually.

11:04:50 From Brian time the loop.

11:05:21 From Jeff Miller Then you add another tennis ball, and it's a crazy but fun, active scene.

11:05:36 From Brian Like adding a metronome for music...it changes everything with some added pressure, but can be very helpful for learning.

11:05:56 From Jeff Miller "Now, optimize!" and we ended up with an emergent pattern of a circle of people handing balls around. And for several PLoP years, then we returned to that game.

11:06:56 From Brian It does "equalize" everyone for a few minutes.

11:07:02 From Jeff Miller Gerald Weinberg was a serious older white man who appreciated games as a learning mode. "tickets to talk"

11:09:53

11:09:53 From Jeff Miller Brian reflects: the thing I appreciate about meetings vs. conferences:
"this is a meeting for efficient resolution of business"
"this is a meeting for relationship building and camaraderie" Ward says: you want that camaraderie, so you know who to depend on when you need help.

11:11:52 From Jeff Miller While getting to know people for how they can be useful may sound instrumental, it's actually a positive-sum strategy; looking after one's own interests can look after everyone's interests.

11:13:53 From Jeff Miller Ward reflects that working with Kent Beck was accidental in a particular way, where the research group needed to justify itself and have a corporate client; it slowed down the research process.

So Ward and Kent did little day-length research projects.

Discovering new ways to deal with software.

Verbalizing experiences and insights, learning and becoming aligned in our points of view.

Having an interactive workstation where you could discuss the software and point to it.

11:14:05 From Marc Pierson “Different but becoming aligned!”

11:14:36 From Brian "Software for your head" is an intersting case study with patterns for thinking... https://www.amazon.com/Software-Your-Head-Protocols-Maintaining/dp/0201604566

11:15:24 From Jeff Miller I thought of Software for Your Head as both intriguing and psychologically unsafe.

11:15:47 From jan dittrich (er/he|they) The brains-run-software is such an interesting metaphor.

11:15:56 From Jeff Miller It seemed very particular to the group that developed it.

11:16:10 From jan dittrich (er/he|they) Reacted to "It seemed very parti..." with 👀 Gigerenzer, Gerd, and Daniel G. Goldstein. 1996. “Mind as Computer: Birth of a Metaphor.” Creativity Research Journal 9 (2–3): 131–44.

11:23:07

11:23:07 From Brian Burros can mess you up too... Bye Marc.

11:24:37 From Paul Rodwell https://archive.org/details/psychologyofcomp0000wein

11:28:44

11:28:44 From jan dittrich (er/he|they) AtlasTi, NVivo, MaxQDA…

11:29:44 From Brian I think part of the "scaling" issues should be to focus companies on the size of ~50 people, and if that doesn't meet demand, then we need more "copies" of the company doing the "same" thing...

11:30:50 From Jeff Miller Cook and Allspaw approaches (LFI, from Eric)

11:31:53 From Jeff Miller Jan discusses tools for qualitative analysis and discussion of text (AtlasTI, NVivo, MaxQDA) from academia which might be interesting to examine as software systems.

11:32:08 From jan dittrich (er/he|they) Krippendorff was important in this "developing testable schemas"

11:32:36 From Jeff Miller Eric says: "what tells you that these indicators are relevant for qualitative analysis? what makes a useful schema relevant and repeatable?" Eric says: Use a rigorous approach: substantiated by evidence present in the conversations. ("Stay close to reality").

11:34:06 From Jeff Miller Eric says: we spend so much time trying to invent the future that we can get misled in our evaluations of things from the past -- not "what happened?" but "how could it have been better?" without strong grounding.

11:35:15 From Jeff Miller Ward discusses risk reduction strategies, where XP is a risk reduction technique; he had a conversation with an entrepreneurial incubator creator -- innovative methods, look at a business plan and look at what they're planning to do.

11:36:29 From Jeff Miller The incubator creator said: "Don't take their view for granted. Start first with the first sentence, and unpack the assumptions that the person is making; cross out any assumption handled later on in the business plan. When the margin fills up, discard the business plan, you don't want to work with that person."

11:36:53 From Brian What are you trying to do? Articulate your objectives using absolutely no jargon. 
How is it done today, and what are the limits of current practice? 
What is new in your approach and why do you think it will be successful? 
Who cares? If you are successful, what difference will it make? 
What are the risks? 
How much will it cost? 
How long will it take? 
What are the mid-term and final “exams” to check for success?

11:37:52 From Jeff Miller Jan describes a method of analyzing a text by putting a narrow window around a sentence and exploring "what does that look like it means now?"

11:41:10

11:41:10 From Jeff Miller From Ward's friend, his framing: "How are we going to make money together?" after more advanced examination of a business plan for viability. He also would sometimes use a more aggressive sort of personal challenge, as a way of evaluating a person's behavior under stress.

11:42:15 From Jeff Miller Weinberg: "Many software managers are still judged largely on deliverables, not on their ability to create teams that can deliver."

11:43:16 From Paul Rodwell The Mythical Man-Month - published 1975

11:43:36 From Jeff Miller Jan asks about Fred Brooks, was he involved in the Patterns community?

Ward observes that Brooks was really before that time, his observations about struggles with OS/360 were later incorporated into IBM's doctrines for federal contracting.

11:44:37 From Paul Rodwell Thinking of people thing - DeMarco and Lister - Peopleware: Productive Projects and Teams

11:45:36 From Jeff Miller yes, I enjoyed Peopleware in the day, though the BIlly Joel song references wore down on me after a while. But it does make a good point about "software teams! they're made out of people!"

11:51:33

11:51:33 From Jeff Miller Ward's comments on C3 and the particular strengths of the people on the team -- Kent Beck, Martin Fowler, Chet Hendrickson, Ron Jeffries -- as a positive synergy both on the team, and in being able to turn the successes of teamwork into a method or group of methods for applying agile development to corporate settings.

11:52:49 From Paul Rodwell back from a rabbit hole looking at some later DeMarco work - a nice section after the bullet list in this review of “Adrenaline Junkies and Template Zombies: Understanding Patterns of Project Behavior” - https://www.academia.edu/55239792/Adrenaline_Junkies_and_Template_Zombies_Understanding_Patterns_of_Project_Behavior

11:53:08 From Pete Chet Hendrickson

11:53:37 From Jeff Miller Kent: can give an answer to crystallize a how or why.
Martin: can explore approaches visually, clarify structure and thoughts
Chet Hendrickson: senior programer, secure, and willing to take the blame so that the team could move on
Ron: had worked in corporate settings, could add context and nuance ty Pete! will fix in post-production

11:54:12 From jan dittrich (er/he|they) (Smalltalk being a "secret sauce"): have seen that described as a (management) mc guffin to get people moving.

11:54:31 From Jeff Miller Ron: "YAGNI", You aren't gonna need it, OOP lets you add the thing later on if you need it, especially Smalltalk.

11:56:04 From Paul Rodwell Replying to "back from a rabbit h..." link to a book preview - https://www.dorsethouse.com/pdf/Preview-Dorset-Adrenaline.pdf interesting list of patterns…

11:56:49 From Brian /s you can turn your CPU into a heater with Java /s

11:58:56 From Jeff Miller (discussion about different approaches and languages and the heritage of influences) Rust
-- how it evolved from the C++ ownership related bugs in Mozilla
-- very strong community around compiler error messages being useful in the developer's context.

12:00:33 From Jeff Miller Eric's reflections on "it's too soon for me to answer that question to the compiler!" about Rust. @Eric Dobbs  - does Rust amplify that sense for you?

12:01:28 From jan dittrich (er/he|they) https://jdittrich.github.io/quickMockup/

12:01:50 From Jeff Miller Having to stop and define a type - is that a useful bump, or not so useful? / vs. Javascript as "an object is a bundle of properties and a bundle of associated functions".

12:02:09 From Brian I went to LanNext.2014 and saw Niko Matsakis give a Rust talk that was pretty interesting. This is the panel with a few other interesting people. https://learn.microsoft.com/en-us/events/lang-next-2014/panel-systems-programming-languages-in-2014-beyond

12:02:10 From Jeff Miller (Ward's reflection on thinking with code, and premature concreteness) Python: variables are untyped, values are typed. (sometimes enough flexibility)

12:04:03 From Jeff Miller Ward describes a filesystem browser;
a directory is a collection of files;
the key "size" is also a number;

Nick says:"use a symbol rather than a string, it's a unique identifier that won't collide with any file name".

12:05:25 From Jeff Miller Ward describes the process of the code shrinking and producing correct but surprising behavior, and so decided that it needed to be described; it was slightly tricky, relying on the Scheme and Self heritages in Javascript.

12:06:07 From Paul Rodwell https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Symbol

12:07:17 From Brian Comments don't bridge across interfaces very well, so where independent code is intented to interact, much better if the constraints can be implied/imposed by the structure.

12:07:32 From Jeff Miller Ward: "I feel as productive in Javascript as I did in Smalltalk; the tools and support infrastructure are strong. However, I have to know about ten times as much."

12:07:58 From Brian I've hit a few cases where people are a bit too cleaver with JSON and it makes it hard for cross platform interoperatbility.

12:08:13 From Jeff Miller Ward: "I'm less pleased with JQuery and Coffeescript than I was when we picked them up."

12:08:42 From Brian JQquery was very bottom up from the trenches.

12:09:09 From Jeff Miller Glamorous Toolkit story: I had a similar point of trouble with losing keypress signals, like Federated Wiki sometimes does with the interaction of JQuery and Leaflet. "put it in the DOM, read it out of the DOM"

12:10:45 From Brian Hence the family of MVC type patterns...

12:11:20 From Jeff Miller jQuery as a library that allows you to step on yourself if you don't have a mental model of what information is in the DOM.

12:12:54 From Jeff Miller Ward relates a story of prototyping Federated Wiki's client in React, and how it was difficult to manage all the relationships needed for interaction. There were learning points from the React prototyping that helped reveal important points for how Federated Wiki works with pages.

12:14:05 From Jeff Miller A page on the disk is different from its representation on the browser UI.

12:14:17 From Brian I'd think that the server page is "synched" with the local storage page in the browser and the panels interact with the local storage and then sync the local storage with the server for permanant persistence.

12:14:32 From Jeff Miller On the client, it's described as a "panel" (pane?)

12:17:16

12:17:16 From Jeff Miller Eric reflects on the Observable approach. "Most things that I want to do with the Federated Wiki are going to be handled as runtime variables, differently from the Observable Notebook approach where the dependencies are established and saved as notebook state."

12:18:56 From Jeff Miller Eric says: My big gripe with React is -- I've told a lot of folks this story before -- we'd give them a computer homework problem, like a chatbot. I tried to install their code, and didn't want to install a huge pile of dependencies on my laptop, so I used a container, and then "npm install" watched a pile of dependencies -- 56,000 or so dependencies in a React chatbot."

12:19:59 From Brian It's in their favor if the "workforce pool" is versed in the enterprise tools...even if it's not the typical real problem.

12:20:14 From Jeff Miller theme:
does this well-regarded system
actually solve the problem you need
with good trade-offs? Brian: yes, it's one of the open source X corporate dynamical system things.

12:20:40 From jan dittrich (er/he|they) Replying to "It's in their favor ..." Yeah, I often wonder how such enterprise problems seep through whole ecosystems.

12:21:01 From Jeff Miller and also fashion driven and resume driven development

12:21:20 From jan dittrich (er/he|they) vue is much nicer than react in the dependency regard. You can still have a vue.js file locally and it does. Did not try out working with observables directly.

12:21:29 From Jeff Miller the new shiny buzzwords arrive faster than we can evaluate them

12:22:03 From jan dittrich (er/he|they) it has gotten much slower though! (at least in JS frontend space)

12:22:03 From Brian For the little web programming I've done, I do appreciate a nice templating framework.

12:22:06 From Jeff Miller Observable can work on an airplane, or work locally, but their business model wants you to use their service online.

12:22:49 From jan dittrich (er/he|they) http://aprt.us/

12:22:59 From Jeff Miller thanks! "apparatus UI framework" was all misses the clean framework advantage for things like Croquet and Observable ?

12:24:23 From Brian TTYL.

12:24:52 From Paul Rodwell Replying to "http://aprt.us/" Toby Schachman

12:26:21 From jan dittrich (er/he|they) vue mastery classes has a nice video course that explains their dependency tracking. (1-2h I think) Replying to "vue mastery classes ..." also I think, nice intro to symbols and intercepting property access.

12:28:07 From Jeff Miller "Build Systems a la Carte" : https://www.microsoft.com/en-us/research/uploads/prod/2018/03/build-systems.pdf Mokhov, Mitchell, Peyton-Jones "Make", "Excel", "Bazel" (Blaze), "Shake"

12:29:10 From Jeff Miller Discord: more fan content Slack: expensive for corporate systems with memory

12:31:25 From jan dittrich (er/he|they) Abbott: "Chaos of Disciplines" (answering the same questions again and again)