CHAOSS Data Science Working Group, April 21, 2026

CHAOSS Data Science Working Group, April 21, 2026 ![rw-book-cover](https://img.youtube.com/vi/eSPM4laf5VI/maxresdefault.jpg) 🔗 **Source:** [CHAOSS Data Science Working Group, April 21, 2026](https://www.youtube.com/watch?v=eSPM4laf5VI) > [!note]+ Document Notes > tags: #chaoss #data-science #open-source #metrics #community #governance #contributors > author: CHAOSS Data Science Working Group > source: YouTube > reference: https://www.youtube.com/watch?v=eSPM4laf5VI > date created: 2024-12-19 10:15 > > # 🏷️ Title: CHAOSS Data Science Working Group, April 21, 2026 >  ## 🧰 Action Inbox - [H1] Set up Docker container with sample Augur data for contributors — next couple weeks #data-science #accessibility - [H2] Create foundation stats project structure and contributor retention strategy — ongoing #foundations #collaboration - [H3] Formalize repository creation process under CHAOSS governance — community call discussion #governance #process ## 🧭 Table of Contents - [[#📝 Summary]] - [[#📖 Main Content]] - [[#🗒️ Notes]] - [[#📚 References]] - [[#🗣️ Paraphrase]] - [[#📰 Newsletter]] - [[#✅ Action Items]] - [[#🧪 Hypotheses]] - [[#📊 Methodology]] - [[#❓ Questions]] - [[#🔮 Future Research]] - [[#💡 Implications]] - [[#🧠 Learning Styles Assessment]] - [[#📊 Accessibility Assessment]] - [[#🌍 DEIB Assessment]] - [[#✅ Logical Validity]] - [[#🛡️ DARVO Analysis]] - [[#🎯 Conclusion]] ### 🔗 References - [CHAOSS Community](https://chaoss.community) - [Augur Documentation](https://oss-augur.readthedocs.io) ## 📝 Summary - CHAOSS Data Science Working Group meeting covering current projects and barriers to contribution - Key focus on foundation stats project, Augur data access, and contributor retention challenges - Discussion of AI impact on data science workflows and limitations - Plans for formalizing repository governance and reducing barriers to entry - Working group evolution toward multiple research sub-projects under umbrella ## 📖 Main Content The CHAOSS Data Science Working Group is undergoing significant changes in structure and focus. With Dawn stepping down as working group lead while remaining involved post-leave, Cali and S have taken over leadership roles. The group has archived the Repel project and moved it under CHAOSS to enable visualization work that builds on CHAOSS metrics models. A major initiative is the foundation stats project led by Ernest, which aims to normalize metadata across different foundations (CNCF, Apache, Eclipse) to enable cross-foundational research. The project faces challenges with contributor retention despite initial interest. The group is working to reduce barriers to entry through improved data access, including plans for Docker containers with sample Augur data. There's ongoing discussion about balancing AI assistance with human analytical work, particularly around messy and unstructured data analysis. ## 🗒️ Notes #### ⏱️ [00:00](https://youtu.be/eSPM4laf5VI?t=0) — [H1] Working Group Leadership Changes - Key idea: Dawn has stepped down as working group lead but remains involved post-leave - Evidence: "Dawn is no longer the head of like the data science role that she was in, but she's still and so she has stepped down as a working group lead for this" #### ⏱️ [02:30](https://youtu.be/eSPM4laf5VI?t=150) — [H2] AI Integration Challenges - Key idea: AI tools help with scripting but struggle with strategic analysis of messy data - Evidence: "AI doesn't do this very well but everyone assumes that it does because it's data and like AI does the data thing" #### ⏱️ [05:15](https://youtu.be/eSPM4laf5VI?t=315) — [H3] Foundation Stats Project Overview - Key idea: Working to normalize metadata across foundations for cross-foundational research - Evidence: "foundations report metadata are very very different although I mean CNCF papache and Eclipse they they record metadata structures in different ways" #### ⏱️ [08:45](https://youtu.be/eSPM4laf5VI?t=525) — [H2] Contributor Retention Challenges - Key idea: Contributors join initially but don't return after first contribution - Evidence: "the issue we opened on the data science working group is attracting contributors but most of those contributors they don't come back" #### ⏱️ [12:20](https://youtu.be/eSPM4laf5VI?t=740) — [H1] Data Access Barrier Solutions - Key idea: Planning Docker containers with sample data to lower entry barriers - Evidence: "trying to get a docker image of docker of of augur. So then um people who are coming in who want to um work on visualizations" #### ⏱️ [18:30](https://youtu.be/eSPM4laf5VI?t=1110) — [H3] Repository Structure Discussions - Key idea: Working group evolving toward multiple sub-projects requiring new organization - Evidence: "there's going to be constantly like like sub projects under this coming in and out. How should we structure this?" #### ⏱️ [25:15](https://youtu.be/eSPM4laf5VI?t=1515) — [H2] Augur Development Status - Key idea: Augur setup has dramatically improved but team resources are stretched - Evidence: "the barrier of entry to like getting Augur set up and using it is pretty much it's it's a whole different world than the last time you since you went on leave" #### ⏱️ [32:45](https://youtu.be/eSPM4laf5VI?t=1965) — [H3] Dynamic Time Warping Algorithm Discussion - Key idea: DTW algorithm suggested for analyzing individual contributor pace patterns - Evidence: "dynamic time warping and it's very good for tracking human behavior against their own behavior over like a week, a month, a year" #### ⏱️ [38:20](https://youtu.be/eSPM4laf5VI?t=2300) — [H2] SciOS Collaboration Opportunity - Key idea: Potential collaboration with scientific repository foundation on data resilience - Evidence: "Jonathan Star, who works with Scios. So that's a open source foundation just working with uh like just scientific repositories" #### ⏱️ [45:10](https://youtu.be/eSPM4laf5VI?t=2710) — [H1] Repository Creation Process - Key idea: Need formal process for creating new repositories under CHAOSS governance - Evidence: "what is the difference between a repository that implies a project like an actual project or a repository that is like a sub portion of the data science working group" #### ⏱️ [52:30](https://youtu.be/eSPM4laf5VI?t=3150) — [H2] CHAOSS Community Growth Challenges - Key idea: Growing community requires more formal processes than previous informal approach - Evidence: "it's growing pains that I think are a sign of the specific like like this where it's like two years ago I would have been so and I still am" ## 📚 References - [CHAOSS Metrics](https://chaoss.community/metrics/) - [Augur Project](https://github.com/chaoss/augur) - [Dynamic Time Warping](https://en.wikipedia.org/wiki/Dynamic_time_warping) ## 🗣️ Paraphrase - The CHAOSS Data Science Working Group is transitioning leadership while expanding its scope to include multiple research sub-projects focused on open source metrics and analysis - There's a strong emphasis on reducing barriers to contribution through better data access and tooling, particularly Docker containers with sample datasets - The group faces ongoing challenges with contributor retention and is exploring both technical solutions (better data access) and community solutions (recognition programs) ## 📰 Newsletter The CHAOSS Data Science Working Group is evolving into a hub for multiple open source research initiatives! 🚀 Key highlights include new foundation stats work normalizing metadata across major foundations, improved Augur data access coming soon, and discussions about AI's role in data analysis. If you're interested in open source metrics and analytics, this community offers great opportunities to contribute to meaningful research. The working group is actively working to lower barriers to entry with better tooling and sample datasets. Resources: [CHAOSS Community](https://chaoss.community/), [Augur Documentation](https://oss-augur.readthedocs.io/), [CHAOSS Metrics](https://chaoss.community/metrics/) ## ✅ Action Items #### DevRel - Create onboarding documentation for new data science contributors - Develop workshops around CHAOSS metrics analysis techniques - Document AI tool integration best practices for data science workflows #### OSPO - Establish cross-foundation metadata standards collaboration - Develop security scanning for research data repositories - Create governance templates for research project management #### Community - Foundation stats project: improve contributor onboarding and retention - Create mentorship programs connecting experienced researchers with newcomers - Establish contributor recognition and badging systems #### WordPress - Apply CHAOSS metrics models to WordPress ecosystem analysis - Explore foundation stats methodology for WordPress.org project data - Document contributor pace analysis techniques for WordPress core ## 🧪 Hypotheses - Providing accessible sample datasets will significantly increase contributor retention in open source analytics projects - Cross-foundation metadata normalization will reveal new insights about open source project sustainability patterns - Dynamic time warping algorithms can effectively model individual contributor behavior patterns for community health metrics - AI tools are most effective for data science infrastructure tasks but less valuable for strategic analysis of complex datasets ### 📊 Methodology - Controlled study comparing contributor retention before/after sample dataset availability - Cross-sectional analysis of metadata structures across CNCF, Apache, and Eclipse foundations - Time series analysis of contributor patterns using DTW algorithms ### ❓ Questions - What metadata normalization standards would be most valuable across different foundation types? - How can we better identify and address the specific skill gaps that prevent data science contributor retention? - What governance structures best support multiple research sub-projects under a single working group? ### 🔮 Future Research - Development of standardized open source project metadata schemas - Investigation of AI tool effectiveness boundaries in data science workflows - Cross-foundation sustainability metric validation studies ### 💡 Implications - Practice: Need for better tooling and data access in open source analytics - Policy: Governance structures must evolve to support growing research communities - Security: Data resilience becomes critical as research projects scale - Education: Contributor onboarding requires both technical and community components - Funding: Research sustainability depends on reducing volunteer burden through automation ### 🧠 Learning Styles Assessment Reading level estimate: Graduate Level (16-18) #### 👀 Visual learners - Present: Discussion of data visualization development for 8knot - Improve: Architecture diagrams of Augur data flows, foundation metadata comparison charts #### 🔊 Aural learners - Present: Video meeting format with discussion - Improve: Audio summaries of key technical concepts, recorded tutorials #### 🗣️ Verbal learners - Present: Extensive discussion of technical terms and concepts - Improve: Glossary of CHAOSS/Augur terminology, analogies for complex algorithms #### 🧩 Kinesthetic learners - Present: Plans for Docker containers and sample datasets - Improve: Step-by-step tutorials, interactive Jupyter notebooks #### 🧠 Logical learners - Present: Discussion of algorithms and systematic approaches - Improve: Decision trees for contributor onboarding, process flowcharts #### 🧑‍🤝‍🧑 Social learners - Present: Collaborative working group discussion format - Improve: Structured peer review processes, mentorship matching #### 🧘‍♀️ Solitary learners - Present: Individual research project opportunities - Improve: Self-paced learning modules, independent project templates #### 🌿 Naturalistic learners - Present: Discussion of ecosystem-level foundation comparisons - Improve: Open source ecosystem health analogies, sustainability frameworks ### 📊 Accessibility Assessment #### Screen reader semantics - Present: Video transcript provided - Improve: Structured headings in transcript, alt text for any visual content #### Keyboard navigation - Present: Standard video controls available - Improve: Timestamp navigation, searchable transcript #### Contrast and readability - Present: Clear audio discussion - Improve: Visual presentation materials with proper contrast #### Text resizing and reflow - Present: Responsive transcript format - Improve: Ensure all supplementary materials scale properly #### Media and captions - Present: Full transcript available - Improve: Closed captions, audio descriptions of visual elements #### Forms and interactions - Present: Standard meeting participation - Improve: Accessible contribution processes, inclusive participation methods ### 🌍 DEIB Assessment #### Cultural breadth - Present: International working group participation - Improve: Multi-timezone meeting scheduling, global case studies #### Inclusive language - Present: Generally professional and welcoming tone - Improve: Plain language summaries, avoid technical jargon barriers #### Neurodiversity support - Present: Structured meeting format with agenda - Improve: Meeting materials provided in advance, multiple participation modes #### Representation - Present: Diverse voices in technical discussion - Improve: Ensure balanced speaking time, inclusive leadership opportunities #### Psychological safety - Present: Supportive tone around challenges and questions - Improve: Clear escalation paths, explicit community guidelines ### ✅ Logical Validity #### Strong reasoning - Deductive strength: Clear logic around contributor barriers and solutions - Inductive strength: Experience-based insights about community growth patterns - Causal reasoning: Explicit connections between data access and contribution patterns - Analogical strength: Effective use of algorithm analogies for problem-solving #### Fallacies - False dilemma: Some discussion presents AI as either fully useful or useless without acknowledging nuanced applications - Anecdotal evidence: Heavy reliance on personal experience without systematic data #### Tone of voice - Overall tone: Collaborative, problem-solving oriented, technically focused but accessible ### 🛡️ DARVO Analysis - Deny: No denials or minimization observed - Attack: No attacks on accusers present - Reverse victim/offender: No victim/offender reversals identified - Emotional cues and rhetoric: Supportive, constructive discussion focused on community improvement ## 🎯 Conclusion The CHAOSS Data Science Working Group is at an inflection point, evolving from a small research group into a platform supporting multiple open source analytics initiatives. Key success factors include reducing technical barriers through better data access, establishing sustainable contributor onboarding processes, and developing governance structures that support growth while maintaining research quality. The group's focus on cross-foundation collaboration and practical tooling positions it well to drive meaningful advances in open source community analytics. ## 🧱 Refactor Appendix ### 📚 Literature Split Pad Purpose: Stage CEE units that become Literature notes. Split level: H3 ### [H1] Docker containers reduce contribution barriers effectively > [!quote] CHAOSS Data Science Working Group, April 21, 2026 via YouTube. Published 2026-04-21. Accessed 2024-12-19. > "trying to get a docker image of docker of of augur. So then um people who are coming in who want to um work on visualizations" Technical infrastructure solutions address major contributor onboarding challenges in open source analytics projects. [view highlight](https://youtu.be/eSPM4laf5VI?t=740#H1) #xref #articleseed ID:rw_chaoss_001 ### [H2] Cross-foundation metadata normalization enables comparative research > [!quote] CHAOSS Data Science Working Group, April 21, 2026 > "foundations report metadata are very very different although I mean CNCF papache and Eclipse they they record metadata structures in different ways" Standardizing data structures across foundations unlocks new research opportunities in open source sustainability. [view highlight](https://youtu.be/eSPM4laf5VI?t=315#H2) #xref ID:rw_chaoss_002 --- ### 🔖 Atomic Split Pad (Zettel Candidates) Purpose: Promote reusable claims into Atomic notes. Split level: H4 #### [Z1] AI tools excel at infrastructure tasks but struggle with strategic analysis - Applies to unstructured data analysis requiring domain context - Pointer to 2:30 discussion of AI limitations - Links: [[AI-Limitations]] [[Data-Science-Workflow]] #### [Z2] Growing communities require formal governance processes - Condition: when moving from small informal groups to larger organizations - Pointer to 52:30 community growth discussion - Links: [[Community-Governance]] [[Scaling-Organizations]] --- ### ✏️ Fleeting Queue Purpose: Capture quick actions for Fleeting notes. - [F1] Investigate Dynamic Time --- ## Full Transcript [00:00] I think that's what I want. I don't [00:02] know. This is my first time having the [00:06] the chaos community um persona. [00:10] I've led enough meetings but had never [00:13] logged in for myself. Um but yeah, [00:16] that's what I'm starting to work on. We [00:18] were working on this with our product [00:19] security last year and then they depp [00:20] prioritized the effort and I was like, [00:22] "Okay, do you" and then like a couple [00:24] weeks ago they're like, "Hey, we have [00:26] this new great idea." And I was like, [00:28] "Yeah, yeah, I know. We've done we did [00:31] this. We've already had these com I'm [00:33] like, "We've already been through this." [00:35] I agree. [00:36] Um, [00:39] we have a team that's looking at that [00:41] more closely. Um, I have like [00:42] consultations with them. Um, I wish they [00:46] would publish their findings more. I [00:48] might poke them on that. um because [00:50] >> this would be a really great place to do [00:51] that and would be very happy to [00:54] collaborate on that type of I mean [00:56] that's what I'm hoping to be able to do [00:59] um as things move forward hopefully with [01:02] the product security side for Red Hat um [01:05] is that we which has kind of been the [01:07] same approach that we've had within our [01:08] team the whole like is we upstream [01:10] everything we can and then cut it off [01:12] where it is like internal like decision [01:15] if it starts to point towards any like [01:18] actual internal decision making and [01:19] stuff like that. Um, [01:23] has Google tried to replace analysis and [01:26] everything with AI yet? Right. You might [01:28] not want to put that on a recording. So, [01:30] that's probably you could say no [01:31] comment. [01:33] >> Um, [01:35] no, it's more like a combo like [01:40] >> what what can I do? What can I be helped [01:42] by? But it turns out a lot of my work is [01:45] still working with like incredibly messy [01:48] data and unstructured data of which AI [01:51] does not do very well with because it [01:53] doesn't like know what it is. It doesn't [01:55] have enough context and to like [01:58] >> spend the time to make it have context [02:02] then I've done all the work already. [02:04] Like I went through and like added 200 [02:06] labels to open-ended questions and then [02:08] it was like okay now I could summarize [02:10] it and it's like yeah I did all the [02:11] work. [02:13] >> Yeah, I could have just done this [02:14] myself. Yeah, I think that's as like a [02:17] whole I'm going to be very curious to [02:19] see where like data science and more [02:22] like um [02:25] what like and pretty much just like [02:27] strategic analysis like an and where [02:28] you're actually using an you're using [02:30] data to like inform decisions or inform [02:33] a per uh a perspective on a situation [02:36] whether that be open source or [02:37] otherwise. I'll be very curious to see [02:39] where this goes in the I don't even know [02:43] what long term means anymore because [02:45] everything changes like every two weeks. [02:47] But let's say [02:48] >> in 2 years because realistically at this [02:51] point [02:53] >> AI doesn't do this very well but [02:55] everyone assumes that it does because [02:57] it's data and like AI does the data [03:00] thing. And so it's [03:04] the amount of times that I'm sitting [03:06] here and I'm like I mean sure you can [03:09] have an AI do that. Um it doesn't mean [03:13] it will be good useful or I mean you can [03:16] that's that's a choice you can make. [03:18] >> I I will say it has helped with like [03:22] just the efficiency of writing scripts [03:25] which is that's what it's good at. So [03:27] it's like fix this query for me. Great. [03:30] now like so I think in that ways it has [03:33] made things faster but I'm like it's not [03:37] doing anything strategic it's just it's [03:39] doing the monkey work which is like can [03:41] you fix this function because I don't [03:42] know why I'm getting an error [03:43] >> great thanks [03:45] >> um yeah I that is probably I talk about [03:48] this a lot the biggest thing I use AI [03:49] for is being like okay can you make a [03:51] query do exactly this and it's like is [03:53] this the most efficient I I hit it with [03:55] the is this the most efficient option on [03:58] repeat and then it's like here's [04:00] considerations. I'm like, "Thank you. [04:01] This would have taken me 30 minutes like [04:03] two years ago, which I so that's a side [04:06] effect. I completely agree." But like [04:08] the amount of times where it's like, [04:09] "Oh, open claw will like be able to [04:11] scrape the internet and tell us what to [04:13] what to do." And I'm like, [04:15] >> "Okay, I guess I have to address [04:19] >> I want to tell you one silly one." And I [04:22] guess I could say this on the record. [04:24] It's not revealing anything surprising [04:25] at Google. Uh, we clearly have internal [04:28] tools that build on our internal data [04:29] sets that could never be externalized, [04:31] but then it has the ability to run [04:33] models on our own stuff. And I asked it [04:35] a question and it literally summarized [04:37] my own document. And I was like, [04:41] I was looking for other sources like [04:44] >> I guess I'm just the only person who has [04:46] the intelligence. [04:49] So that was like a a funny moment where [04:52] we're supposed to like give them [04:53] feedback and I was like asking a [04:54] question getting your own data back. [04:58] >> Okay. Um, but yeah, since we only have [05:01] like one or two discussion points, I [05:03] actually I think it's a good time to [05:06] just kind of say like what's been going [05:07] on in this working group because it [05:08] actually has changed a decent like this [05:11] is this is probably pretty different [05:13] like makeup what we're working on and [05:15] stuff is pretty different than when you [05:18] were last here. Um, and so Don I I'm [05:22] assuming like Don's out for like two [05:24] months do like I'm so happy for her. I'm [05:27] so happy she's not here dealing with all [05:30] this stuff, even though I don't want to [05:31] be dealing with the overarching things [05:34] in chaos right now. But um so Dawn is no [05:39] longer the head of like the data science [05:41] role that she was in, but she's still [05:43] and so she has stepped down as a working [05:46] group lead for this, but is once she's [05:49] off leave will be still involved in like [05:52] the working group. So, it's myself and S [05:56] are the working group leads. Have you [05:59] met S before? Or because I don't know [06:01] S's time. Okay. I was like I I didn't [06:03] remember where the timelines crossed [06:05] over or not. [06:07] >> I've actually I met I've met S in other [06:08] contexts. So, not actually in this [06:10] working group, but we've had [06:12] interactions before. [06:13] >> Nice. So, yeah, S is the other working [06:16] group lead. Um, and I would say like the [06:19] main things that are going on in this [06:21] working group at the exact moment is one [06:23] we've been like I we've archived like [06:27] repel which was like the open research [06:28] that was under like Aspen and moved [06:31] everything over to here. So then people [06:33] can be working on building um like kind [06:37] of the visualizations that would go into [06:40] eight knot that a lot of times are being [06:41] just are based off of the chaos metrics [06:45] models. So actually taking a lot of the [06:46] metric models that are more have stayed [06:50] in a con like a conceptual or conceptual [06:52] and bring them to life. And so those are [06:55] we have all of the notebooks that are [06:56] created for that. Um and so they're all [06:59] based off of I was I've been using um [07:02] data collection tool is my um my neutral [07:06] term for like we're in the while we're [07:08] in this in between period between augur [07:10] and what will eventually come after. But [07:14] so we'll connect to the data collection [07:16] tool that's run by chaos and so people [07:18] can build those visualizations here and [07:21] then we would integrate them into 8 as [07:23] well. So that's some of the work that's [07:25] being done um right now. I can let [07:28] Ernest talk about the um foundation [07:31] stats work. [07:34] >> All right. Thank you Ki. Um you know [07:37] when I joined you working I think that [07:39] should be just be after you left you [07:42] went on the holiday. So um I started [07:45] working with SA to understand um you [07:49] know currently the way foundations [07:51] report metadata are very very different [07:54] although I mean CNCF papache and Eclipse [07:57] they they record metadata structures in [08:00] different ways and it makes it [08:02] impossible for researchers or committee [08:04] managers to answer sambly questions on a [08:07] cross foundational level. So what we are [08:09] trying to do with the foundational stat [08:11] project is to get the meta data together [08:14] and normalize them so that we could do [08:17] research that could give us insight on a [08:20] cross foundational level. That is how [08:22] does the way CNCF report their maybe [08:26] documentation [08:27] how does it impact some sustainability [08:29] metrics as compared to Apache and any [08:33] other foundations. So we've made some [08:36] trials to try to automate the work with [08:39] some AI crawlers but we were not able to [08:41] validate most of our work. So we had to [08:44] revert to doing things manually which [08:46] we've been in that phase for a while [08:48] right now and we are trying to find a [08:50] way to scale that through. So um the [08:53] Apache data collection is still ongoing [08:56] but um I believe that we are putting [08:59] systems in place to make things to [09:01] attract more collaborators. So one thing [09:02] we notice is the issue we opened on the [09:06] data science working group is attracting [09:08] contributors but most of those [09:10] contributors they don't come back. [09:13] Yeah. So those are the those are the [09:15] stages that we are I'm currently [09:16] thinking of how to you know bring them [09:20] coming back you know I'm going to [09:22] discuss that maybe later today then we [09:24] could probably share opinions on how [09:27] that could be solved and all. So that's [09:29] pretty much it and um we are happy back. [09:35] >> Thanks for that. I'm curious if you [09:37] where I could catch up on it. So should [09:38] I go to the repo to see more about this [09:40] project? [09:42] >> Yes, you can. Then we could, you know, [09:44] catch up on Slack afterwards. [09:48] >> Okay. I have to warn you, my Slack [09:51] is like mildly unaccessible at the [09:54] moment. as in like [09:58] >> I can get in but like I can't stay [10:00] logged in. Um and that's a problem of my [10:04] court machine and I I'm not a big fan of [10:06] using two laptops at work right now. So [10:09] I'm trying to fix that, but I like try [10:11] to go into Slack like at least a few [10:13] times a week so I'm not like letting [10:15] things go. But if I'm slow to respond, [10:17] it's because I forgot to log in this [10:19] week. Um but basically like [10:22] >> I yeah it's it's something is going on [10:24] with our network and it it has an [10:27] override for Slack at the moment and so [10:29] I can only log in incognito so I can't [10:31] stay logged in. Yeah, it's a personal [10:35] problem. [10:36] >> We could use GitHub issues, you know, [10:38] commenting on GitHub issues. I I'm [10:42] always very responsive unless if I see [10:44] the message is not does not require [10:46] audience response, but I'm always very [10:47] responsive on GitHub issues as well, [10:49] then we could see how we go from there. [10:52] >> Yeah, there is. [10:53] >> Thank you. Yeah, thanks for sharing. Am [10:55] I I don't know if it's going to be [10:56] helpful or more confusing, but there is [10:58] an is there's a PR that's open right now [11:00] that is like kind of bringing all the [11:02] sources for this project together and [11:04] bringing it under like an like a a a [11:07] name. Um and so it might be a good point [11:10] to like where everything is at, what the [11:13] definition of the project is because I [11:15] think that's um to kind of formalize it [11:18] a bit more. And I think there's there is [11:20] like a kind of a more broad discussion I [11:24] guess for I'm I'm really liking how the [11:26] data science working group is like [11:28] shaping up to be but like okay there's [11:30] going to be constantly like like sub [11:33] projects under this coming in and out. [11:35] How should we structure this? How like [11:37] how do we structure the repository so it [11:40] makes sense to people to go into [11:41] different the different areas? Um which [11:44] I'm I said I'm really happy that this is [11:46] how it's turning out to be. Um but how [11:48] do we how do we handle that? Um cuz I [11:52] can see that there is now like a new [11:54] data set that is around that's really [11:56] interesting around like corporate um [11:58] investment or involvement in open source [12:00] and like I could see an entire project [12:02] coming out around that and like like if [12:05] we're talking about like upstream [12:07] security threat like I see that to be [12:10] the life cycle of this group is [12:13] different research areas coming in and [12:15] out and so how do we want to do [12:23] That's an open question. I'm curious, [12:24] too. I think [12:25] >> we we'll just figure it out as we had we [12:28] grow arms. [12:30] >> Yes. Um [12:32] so, yeah, that's what's been kind of [12:34] that has been the I guess like really [12:37] what's been going on since I guess [12:39] there's been a lot of practitioner [12:40] guides kind of coming in and out. Um I'm [12:42] hoping to work with Dawn on um some [12:46] practition a practitioner guide on usage [12:48] and adoption. Um that's still kind of an [12:52] open open question. Um so then we don't [12:55] just tell people no don't use please [12:57] don't use downloads downloads are [12:59] misleading but a way to be like don't do [13:02] that but here are some ways of going [13:04] about doing this to get this [13:05] information. Um, I had a very funny [13:08] conversation [13:10] in my own like work context where [13:12] somebody was like, "Oh my gosh, stars [13:14] are misleading. What are we gonna do?" [13:16] And I was like, "Yeah, [13:19] yeah, that's nothing actually." Um, [13:23] that's we we knew that which was very [13:26] funny to me. Um, I don't know if the [13:29] person who was brought it up appreciate. [13:31] I was just like, "Yeah, this is this is [13:32] the same the same as before, same as [13:35] after. This is not surprising. Um, but [13:38] it kind of leads to like usage in a do [13:40] like stars, downloads, these metrics [13:42] that people really like to look at that [13:45] don't mean a whole lot. Um, I don't [13:47] think it's sticking to people that those [13:48] are bad because we don't have anything [13:49] to tell them to do differently. Um, and [13:52] so I still find that to be very [13:53] important, but I need a um absolute [13:56] dumpster fire disaster to stop happening [13:58] like every two to three weeks in my [13:59] professional life and then we can do [14:01] that. [14:03] So [14:05] um yeah, but there some of those other [14:06] stuff. We're also talking about trying [14:08] to get a I guess I could just use augur [14:11] for now as the name of like a docker [14:12] image of docker of of augur. So then um [14:15] people who are coming in who want to um [14:18] work on visualizations want to work on [14:20] like 8 would be in a different context [14:22] but starting to develop out any models [14:25] or research on a smaller data set that [14:27] that is available to them. So it is [14:30] something that people can use as like a [14:32] sandbox environment um if they don't [14:34] want to you know set up an entire [14:36] database on their own computer and go [14:39] through all that effort. Want to try to [14:41] get that um barrier of entry of doing [14:44] analysis work as low as possible. Um [14:48] because I do think that's like how do we [14:49] get retain contributors and like that's [14:51] that is my point of view on it. I saw [14:53] someone added that. My point of view is [14:55] how do we make the barrier to entry as [14:58] low as possible? And I think that's [14:59] something that the foundation stats [15:01] portion of the project is a pretty good [15:02] like entryway point to that and that's [15:04] what I'm hoping to have for like the [15:09] like what whether however you want to [15:10] call it the visualization side of it. [15:12] But I see that like expanding more to [15:15] like okay when we're developing I'm [15:16] hoping to work on like security modeling [15:18] of flagging like um upstream risks like [15:22] my hope would be that is something that [15:24] somebody could work on with this [15:26] solution. It might not be the most [15:28] mature option and like provide the best [15:30] results but it is has a has the bare [15:34] bones of all the different data types in [15:37] there so you can start working on [15:39] prototypes and stuff like that. [15:42] I mean, I love that. I think the hardest [15:44] part of our job is always getting the [15:46] data. And so, like, [15:47] >> yeah, [15:48] >> you can't do anything without it. And if [15:50] we have either like existing data sets [15:52] or easy ways to get people access to it, [15:56] like I mean, I would love that [15:58] personally. I [16:00] >> like I'm working on a proof of concept [16:02] right now using my own GitHub personal [16:04] token and it's terrible. [16:07] >> Like, no one no one should have to do [16:09] this. [16:09] >> Wait. Oh, Augur became like, and this is [16:12] why this is all all the things that are [16:14] going on is so frustrating. Um, the [16:16] barrier of entry to like getting Augur [16:18] set up and using it is pretty much it's [16:22] it's a whole different world than the [16:24] last time you since you went on leave, [16:26] which is really exciting. Um, yeah, the [16:28] like but also that's that is a part of [16:30] the I'll actually probably let Adrian [16:32] talk about this because this is more on [16:34] like Adrian's side also. Yeah. I don't [16:36] know if have you've met um Adrian or [16:38] Kyle or um like um these they're both on [16:42] my team now. Um [16:44] >> they were just joining when I was about [16:46] to leave. So like I think we've met in [16:48] passing so I don't really know them. [16:50] >> Yes, we can probably do some [16:52] introductions but that's part of it. And [16:53] as well as there's some different data [16:55] sources that people have just like [16:57] donated to the working group um which [17:00] are really interesting and they're like [17:02] we don't know what to do with this but [17:03] here you go. And we're like, "Cool, [17:06] great. Sounds awesome to me. Can you [17:07] just like The only problem is is I don't [17:09] think people are giving enough context [17:11] of like how did you get this data? Where [17:14] is this data coming from? How is it like [17:16] the a little bit more context so [17:19] something can be done with it?" But [17:20] still like we're moving along. But yeah, [17:22] I'll let Adrian kind of he can do an [17:24] introduction and talk about where the [17:26] like how do we get a access point to a [17:30] bare metals um uh or bare bones auger [17:35] data. [17:38] You we can't hear you. [17:41] >> Okay, that's interesting. [17:44] >> It's now we can kind of hear you but [17:46] it's very low. [17:47] >> Does it work now? It is very It's like [17:50] you're whispering. [17:53] >> How about now? [17:54] >> There we go. [17:56] >> Cool. Jank setups while working [17:58] remotely. Um, hi, I'm Adrian. Um, I [18:02] joined Red Hat last summer. Um, I've [18:04] been I've been basically maintaining uh [18:06] the awkward code base that since then. [18:10] Um, been doing a lot of just general [18:13] improvements and things. um currently [18:15] working through [18:17] events that are happening but um [18:20] definitely going to be continuing to [18:21] work on kind of improving data [18:23] collection and things. Um I think I [18:25] mentioned in the last week's meeting or [18:27] last data two weeks ago um about trying [18:31] to get kind of way like ways to get the [18:35] community access to like just a quick [18:38] like a data set or um instance that you [18:41] could that could just be kind of spun [18:42] up. you don't have to worry about like [18:44] running your own augur, collecting your [18:46] own data. And so I think I think in the [18:49] notes I pointed to a empty I made I I [18:53] think I had previously made a docker [18:54] container that just contains an empty [18:56] database. Um oh no that's for that's for [19:00] um is that still relevant? Do you think [19:03] that's relevant? I mean I don't think an [19:06] empty database is we just need a like or [19:09] a database with a couple of like couple [19:10] of repositories that have been collected [19:12] on is like what is I feel like is the [19:15] kind of pretty much what you get if you [19:17] just start up Augur um out of the box. [19:23] >> Yeah. So that's that's something I do [19:24] plan to get set up uh hopefully along [19:29] with most of the other stuff that will [19:31] hopefully get out soon. [19:33] We'll see how much I have time for. Um, [19:35] but yeah, that's definitely I I do want [19:36] to kind of get [19:39] those like get get a data set of just [19:42] like basic collected data so that analys [19:46] like people making analysis and things [19:48] from that can at least kind of practice [19:51] and see see what it's like to to query a [19:54] loger query the data schema. [20:00] >> Yeah. So, so Kalian Adrian, so it's [20:03] practically impossible to even um work [20:07] on issues right now. Am I correct? [20:10] Because we don't have a sample. [20:13] >> You'd have to Yeah, you'd have to um set [20:17] up augur as well. um which is definitely [20:20] a lot less than what it used to be, but [20:21] you just have to like like run the [20:24] Docker container for then it would [20:26] populate because once you just start up [20:27] um Augur, it populates with a couple of [20:30] different repositories and then that's [20:32] what you would need to work on at um but [20:37] I think it's probably worth the [20:39] discussion of seeing what like yeah I [20:43] know Adrian there's a lot of different [20:44] stuff on the on the plate but seeing if [20:46] of how we can probably like she see can [20:49] try to prioritize getting uh small data [20:53] like that solution handled especially [20:55] with and I'm not these recordings don't [20:58] come out quickly enough slash I don't [21:00] really care for we're planning on [21:02] donating 8 knot formally to chaos and so [21:06] it would be great to have something like [21:08] that with the donation so then if people [21:10] want to start contributing stuff like [21:12] that then the barrier of entry is lower [21:14] but um time is difficult to the exact [21:19] moment whenever there's so much stuff [21:21] going on. [21:23] >> Yeah. And I think actually one of my [21:25] plans in terms of the the data [21:27] collection tooling is like it's always [21:30] been to kind of implement more testing [21:31] and more sort of automatic like hey [21:33] let's make sure that if someone's [21:35] contributing to it that the changes [21:37] actually work. And one of the things I [21:39] like I I think I see a really neat [21:41] overlap between if we could if I can [21:44] kind of maybe not the initial runs but [21:47] in in future like get a a sort of [21:50] automatic process running that would [21:52] basically run through a small subset of [21:54] repos and fully collect them and like [21:58] keep a keep a Docker image up to date [22:00] with that data so that like we're [22:02] talking about to to provide that for for [22:04] people working on that. Um that would [22:06] also be really helpful for for the data [22:09] collection tooling and and the testing [22:10] of that so that we can basically say hey [22:12] okay after a full collection has run [22:14] let's like do some database queries and [22:17] make sure that the data we expect to be [22:19] there is actually there. Uh, so I think [22:22] in the future that's going to become [22:23] very much part of the the sort of [22:25] testing system and the the neat side [22:28] effect is there's going to be an [22:29] up-to-date container with at least some [22:32] sample data in it for people to play [22:34] around with. [22:45] >> Thanks for sharing that. I'm ready to [22:47] test it when you're when you're there. [22:50] Yeah, I was the one that added the [22:54] question of how do you retain [22:55] contributors? Although I'm not I'm not [22:57] talking broadly. I'm not talking based [22:59] on the the issues that I think we are [23:01] facing uh on the foundational staff [23:04] project. So uh I think I read a couple [23:06] of papers I can't remember a particular [23:09] paper maybe a couple of months ago or I [23:12] saw or I saw a project from chaos that [23:15] say there is a problem of non-code [23:18] contributors you know it gets difficult [23:20] for people to know what they've done in [23:23] those communities and one of those ways [23:25] to appreciate non-code contributors is [23:27] to give them maybe a a badge or um a [23:31] weekly or any kind of periodic [23:33] appreciation on socials. [23:35] Maybe that could be some sort of reward [23:37] system for them. Then while I was [23:40] critically thinking about how do we [23:41] because for me um I don't have the [23:44] luxury of time to be doing the manual [23:45] collection myself even though I wish I [23:48] could but um I'm happy to see a lot of [23:51] people do that and let's just do things [23:53] in paralle and see how we could go that [23:56] far. So the question of how do I really [23:58] sustain a a couple of people that have [24:01] come to that repo and contributed then I [24:04] thought of this I wanted to discuss with [24:05] everyone to peruse that and see if we [24:08] could try that out you know I'm happy to [24:10] design a program more formally [24:15] >> my personal I guess perspective of this [24:17] is that making [24:21] I guess like putting towards efforts of [24:23] like how do we make data sources [24:24] accessible like how do we have like [24:26] projects for people to work on. That's [24:28] where my I guess that's where I would [24:30] think the efforts around how do we [24:32] attract and retain contributors? It's [24:35] how do we have meaningful work for [24:37] people to be able to do because a lot of [24:39] times we do have a lot of people who [24:41] comment on um in the data science [24:44] working group in different context of [24:46] like looking for [24:48] um yeah looking to do data science work. [24:50] And so I guess my personal perspective [24:53] is like instead of we we only have like [24:55] a very small set of people instead of [24:57] trying to put time towards like a [25:00] badging effort, maybe putting time [25:01] towards like trying maybe like we can [25:04] figure out how to help Adrian like get [25:05] the data set together for them people [25:07] can start to contribute more to this [25:09] data set or to the data science working [25:12] group in 8 or how do we get the [25:14] foundation stats in a situation of like [25:16] how do you use the foundation stats or [25:18] like how what type of research do we [25:20] want to do? So, that might not be the [25:22] answer you're looking for, but that's [25:24] just my I guess my personal perspective [25:26] on it of like where I would want like [25:29] time investment within like this group [25:31] to go. [25:41] Yeah, those sounds good on Yeah, on a [25:45] broader sense. And I don't know if [25:47] they're out of Okay, Sophia, [25:50] >> I'm just kind of curious because I have [25:51] been out of the loop for a bit, but just [25:54] in terms of the folks that have come in [25:56] and out of the group, what kind of [26:00] backgrounds do they have? Because I [26:01] guess the hard part about like getting [26:04] them involved in a project is sort of [26:05] like [26:07] h knowing that everyone's experience [26:09] level is all over the place um in terms [26:11] of what they've done already and what [26:13] they would want to take on. And I feel [26:15] like there is a high barrier to entry [26:18] for something like data science [26:19] especially if it is the data access [26:21] point and then it's learning the tools [26:23] to interact with it and then it's doing [26:25] the analysis on top of it and then is [26:27] making sense of it like there's there's [26:28] many layers where I'm curious what [26:32] people are what back if we have a sense [26:34] of what skill sets people have when they [26:36] were coming into the group and that [26:38] could help maybe tailor more of an [26:40] experience that could keep them involved [26:42] or get them involved early. I'm not [26:44] really sure what that is. So I I I don't [26:45] know how to advise without knowing the [26:48] folks that have been showing up. [26:53] >> I think I have my starting point. I will [26:55] try to know them because I also don't [26:57] know them. Especially for those that [26:59] have been contributing to foundational [27:00] stats. I just see them on GitHub. They [27:03] just send their p request and that's it. [27:05] I think that should be the first [27:06] starting point for me. I will try to [27:08] care to know them because what you said [27:11] right now I've tried to remember what [27:13] made me like still contributing to chaos [27:16] is because yeah align perfectly with my [27:19] research interest you know I found a [27:21] project and I just I just stuck with it. [27:24] Yeah. So I will try to know them better [27:27] from this coming week. Maybe I could get [27:29] them to start coming to this meeting. [27:32] Then from there we could fix the other [27:35] um you know infrastructures around how [27:39] to give them work in terms of projects [27:40] that align with their interest. [27:48] Um [27:50] if I could ask can hear me please. [27:54] >> Sorry. [27:56] I said if I may ask, do you have a [27:59] timeline for the [28:01] um the sample ago docker that you said [28:05] you have on your um radar? [28:09] >> Uh not I think my main focus right now [28:12] is trying to get a repository out there [28:15] for the for the kind of a community to [28:17] continue contributing under under chaos. [28:19] I think I I am going to probably want to [28:22] do that uh get sample data either [28:25] alongside that or soon after um is is my [28:30] hope. So, [28:32] >> okay. Okay. Thank you. [28:35] >> I would think so like the next couple of [28:37] weeks you think is a fair like [28:39] assessment? [28:41] >> I think so. Yeah. [28:42] >> Okay. [28:45] Yeah, definitely. I mean, it's just [28:48] being reasonable. Like, our team is so [28:51] deeply impacted by the move by Augur [28:53] that pretty much anything in every [28:55] context that we said that we were going [28:57] to do before that has happened [29:01] dramatically shifted. I'm hoping that [29:02] time is coming to a close soon. Um, but [29:06] like we're pretty upfront that that's [29:10] been unfortunately a huge time like [29:13] where almost all of our time has had to [29:15] go when it comes to our upstream [29:16] involvement over the last month. [30:01] I have a random question. [30:03] >> Yeah, if we're in that in that time of [30:05] the meeting. Um, I have started to also [30:09] pop in a bit on the package metadata [30:12] working group. Um, I'm curious if you [30:17] see any alignment between the things [30:19] that they're working on because I know [30:21] >> at least from from Andrew's perspective, [30:23] like I know he's done a lot of work to [30:25] create data sets. Granted, his are very [30:27] focused on package managers. Um, [30:31] but I there could be some alignment [30:34] between the groups. I don't want to pose [30:35] it yet because I I want to get better [30:37] understand what they're working on and [30:38] how it potentially could support this [30:40] group. Um but I'm kind of thinking about [30:42] that as sort of the longtail as longtail [30:45] support. Um if that group is working on [30:47] making data sets more accessible for [30:50] package metadata, then that's also [30:53] potentially [30:54] something that we could point to as part [30:56] of this working group as a sample to [30:58] start with. Um, right now they're very [31:00] much focused on like metadata alignment [31:04] um, and trying to essentially improve [31:07] how package managers expose this [31:09] information. So, it's sort of like fix [31:12] fix the underlying infrastructure. I [31:14] don't know how much influence they'll [31:15] have. U, but it's still an interesting [31:16] prospect of trying to create alignment [31:19] across I don't know hundreds of package [31:21] managers. Um and so like solving some of [31:24] that data integration challenge that we [31:28] face within any data science project. So [31:31] I just wanted to to float that I am [31:33] sitting in on those calls now. Um, I [31:36] won't bring anything to the meeting [31:37] unless it's I think relevant to this [31:38] group. But I was curious if you, Cali, [31:41] or others in the group have already been [31:44] in touch with that group or if you had [31:46] similar thoughts around how and when to [31:49] not say join the parties, but like [31:51] interact with them. [31:54] Yeah, I honestly don't know much about [31:56] it to the point that I was what I was [31:57] typing out was it would be great for [32:00] somebody from that group to come like I [32:01] would be very like open to the idea of [32:03] using one of these like uh meetings or [32:05] at least part of the meetings for [32:07] somebody from that group to present on [32:08] what they are working on um even not [32:11] even from a wanting to establish [32:14] collaboration at that point for for us [32:15] to at least know what they are doing [32:18] where they're going what's the format of [32:19] the data the goals I think that would be [32:23] helpful at least to me because I've [32:25] known I've know of them by name but [32:28] that's pretty much where my um knowledge [32:31] of the group has been [32:34] has stayed. [32:38] >> Sorry. Is the group within chaos? [32:42] >> Oh, and is it related to the [32:44] >> to the URL you sent earlier, Sophia? [32:48] >> Um no, it's a different one. I I'll send [32:50] you Andrew's link. So you have full [32:52] context. [32:54] >> Okay. Okay. Thank you. [33:13] Um, bye y'all. [33:21] Hi. Uh, [33:24] >> hello. Sorry I'm late, but uh, I had to [33:26] go see another mente's talk. It was [33:29] about security in Africa, and it did [33:30] make me cry. It was a really good talk. [33:35] >> Yeah. [33:40] >> So, what are we in what are we covering? [33:43] >> We were just talking about the package, [33:44] man. like we've just kind of honestly [33:46] been throwing around different topics of [33:48] um getting Sophie up to date of what [33:51] we've worked on. We've talked about [33:53] honestly just a bunch of random topics. [33:55] The only thing that I was going to kind [33:56] of tee up that um was on the agenda um [34:00] was a reference to a thread that Josh [34:03] Burkus had opened this past week. Um and [34:08] it was like a discussions around what [34:11] are some approaches for analyzing the [34:14] pace of an individual contributor's like [34:17] contributions. Um was trying to figure [34:20] out like trying to develop out the [34:23] thread on the screen if that's helpful [34:26] for people. Um even though he is not [34:28] here necessarily, I told him that I [34:30] would like kind of tee up the [34:32] conversation in case people had ideas [34:35] about um like I kind of I'll just read [34:39] the interesting analyzing the pace of [34:41] individual contributors contributions [34:43] that is on a smooth frequency basis how [34:46] often do they contribute this frequently [34:49] stats can be used for a number of [34:51] community measurements purposes. [34:53] However, I need an algorithm that gives [34:55] me an adjusted spacing weighted for [34:58] different contributions in smooth where [35:01] it does not get thrown off like things [35:03] like weekends and vacations. Um, pretty [35:06] much she's like while I can write [35:07] something like this from from scratch, I [35:10] seems like there would already be an [35:12] existent algorithm library somewhere [35:14] that has this type of function. So I [35:16] thought this was like an interesting [35:17] topic and kind of some of the [35:19] discussions I'd like to see in like in [35:21] the data science working group of okay [35:23] what are what out of their know people's [35:25] knowledge of different algorithms like [35:27] does this use case people have anything [35:29] that this matches that that seems to be [35:32] a good fit to um model this type of [35:35] activity um if people have any thoughts [35:38] about it [35:42] >> dtw [35:43] dynamic time warping [35:46] Okay, dynamic time DT. [35:49] >> It's the coolest sounding, [35:52] >> stupidest, simplest algorithm. Um, but [35:56] uh what it allows you to see, it's often [35:58] used as an anti- fraud. So like in the [36:01] early days of like seeing how someone [36:02] types into forms to make sure it's them. [36:05] Um, that's dynamic time warping and it's [36:07] very good for tracking human behavior [36:10] against their own behavior over like a [36:13] week, a month, a year. [36:16] >> Sweet. I just put that in the chat if [36:18] you have any um like links or anything [36:21] that refer to um I don't know if you've [36:24] ever worked with Josh Burkus directly. [36:26] Um he's on our like on in Red Hat's [36:30] what's now called Oypo which is just our [36:33] open or OSPO [36:35] um team working he does mainly stuff in [36:38] the CNC he's deep in the CSC CNCF land [36:42] but has a pretty deep data background [36:45] one of the original like people working [36:47] on Postgress um so he kind of hops in [36:51] and out um with working on stuff and um [36:55] always likes to bring in some [36:57] interesting topics. [37:02] >> Yeah. Ernest is asking if it's my [37:04] favorite algorithm of all because before [37:06] I was willing to take on or I'm ever [37:08] willing to take on someone for a data [37:09] science mentee, I asked them to tell me [37:11] what they would apply this algorithm to [37:13] because it is my favorite. But um yeah, [37:17] I really think this might be a good use [37:19] case because it's very good for tracking [37:20] like individual behavior within a crowd [37:23] essentially. [37:25] Um, but you have to have the element of [37:27] time. And so if you have time stamp, [37:28] which we do in this case, then it's a [37:30] very relevant algorithm. [37:33] Plus, it's my favorite. So if it fails, [37:35] that's fine. But it's always fun to play [37:36] with. [37:40] Um, I did want to talk about I haven't [37:43] really gotten to digest the notes for a [37:44] minute, but I just had a call on sort of [37:48] like connecting about chaos to Jonathan [37:51] Star, who works with Scios. So that's a [37:55] open source foundation just working with [37:58] uh like just scientific repositories. [38:02] Um so it was a good conversation because [38:05] I was like where do you get your money [38:06] from? He's like we just reached out to a [38:08] bunch of people who made a lot of money [38:10] with open source but specifically in the [38:12] ways where like a biochem lab uses a lot [38:14] of open source but they make billions [38:16] right so they've got money to give away. [38:17] Really interesting paradigm shifting. [38:19] So, I wanted to talk to him about the [38:22] uh like sponsorship model and if it was [38:25] something they'd be willing to adopt. He [38:27] said he was curious and he said his [38:30] biggest concern with most data science [38:33] projects is that they don't have very [38:35] good um [38:38] like access to like resilient data [38:41] storage is what he wants to work on [38:42] which I thought was really interesting. [38:44] So, it might be a little bit outside of [38:45] the scope of chaos, but there's also [38:47] going to be meetings around that space. [38:50] So, I think we should show up. I'm [38:52] willing to show up and just say like, [38:53] "Hey, there might be something to uh [38:56] co-work here with." And they're a nice [38:58] little nonprofit focused on open source. [39:01] >> Yeah. I would be So, I've talked to [39:03] Jonathan Star before. I would be very [39:05] curious to know how this work does or [39:08] doesn't connect to what he was doing [39:10] like two years ago because I would be [39:12] like I'm like is this a new because it's [39:13] like vaguely similar. It's one of those [39:16] just being I'm like is this the same [39:18] thing under a different name or is this [39:19] something kind of related but but like a [39:22] whole different space because that would [39:24] be a really helpful um to like advise on [39:28] where cuz he's he's been in this I mean [39:29] like I can like kind of like all of us [39:30] is like he's been in this realm for a [39:33] while and so I just don't know if what [39:36] what this is in relationship to what he [39:38] was doing two years ago um would be [39:40] helpful context. [39:42] >> I have no idea. So, I just called him to [39:44] talk about this sponsorship thing and I [39:46] today was the first time I've talked to [39:48] him. So, [39:49] >> yeah. Um cuz he has come he might have [39:52] come to this working group actually and [39:54] spoke before. I could be wrong. He's [39:56] definitely come and spoken at a chaos [39:59] working group that I was attending here [40:03] before and it would make the most sense [40:05] that it would be this one, but that's [40:08] not a full confirmation. [40:13] Well, yeah. Um, I don't know. So, I'm [40:16] gonna see especially I'm going to just [40:18] show up out of my own interest for [40:20] Gadfly to the talks around like data [40:22] resilience. Um, [40:24] >> yes. [40:25] >> Yeah. Because we've got a few projects [40:27] that might turn into large data [40:29] projects. So, [40:30] >> got to be careful. [40:31] >> Yeah. [40:33] um that it kind of ties in well to what [40:36] we were some of the stuff that we were [40:37] talking about um before you had hopped [40:40] on the call of how do we like we're now [40:43] starting to gather some data sources. [40:45] How do we make it to where there are [40:46] consistent ones that people can use for [40:48] different use cases? So this all fits in [40:51] pretty well with what um we were talking [40:54] about before cuz it is [40:57] very interesting that there the problems [40:59] that we have in data science are still [41:01] kind of the same ones that were 5 years [41:03] ago but people think that they have [41:06] dramatically changed because like AI is [41:09] a like because a or LLMs are like the [41:12] thing that everyone thinks about. [41:13] They're like, "Oh, I can just ask this [41:14] question. So, this means that all of the [41:16] data problems are solved and handled." [41:18] And it's like, "No, [41:20] that's not how that works." [41:27] Well, yeah. I don't know. For Godfly, we [41:30] have like triple backup for our data [41:32] projects and I've got a nice little [41:34] server and I know exactly where it is, [41:35] right? But that's a very different one [41:37] >> from most people who have it sorely on [41:41] their computer or on some platform that [41:43] they can't guarantee. [41:44] >> So [41:46] >> but that's all right. But yeah, if [41:48] anything interesting comes out of that, [41:49] I'll come back here and just uh hey, [41:51] here's some advisements around these [41:52] kinds of projects. [41:54] >> Yeah, that'd be great. [41:56] Um, we have about four more minutes. I [42:00] don't know if anybody has anything else [42:02] that they would like to [42:05] um Yep. talk about. [42:08] >> Yeah, I was going to ask S what is the [42:11] next plan for the Oscos migration? You [42:15] know, we had I remember Khali open to [42:20] asking that what feedback do we needed [42:23] from one and I could not necessarily say [42:26] this is what we needed to review. to me [42:29] the um current plan looks good you know [42:33] and I wanted to know what's the next [42:36] step for us to move it to a new repo. [42:38] Yeah, I think what we should do is just [42:41] jump on uh and like look at the code and [42:44] repo tandem for about 45 minutes. And in [42:48] that, the reason why this migration's a [42:50] little bit complicated is when I tried [42:51] to set up the new repo, I broke a bunch [42:53] of the pipelines. Go me. Um so I need to [42:58] think a little bit about parsing that [42:59] out and just doing a merge that is not [43:02] uh broken. [43:05] So, if we can get like 45 minutes this [43:07] next week or so, we should be able to [43:09] clear that up and then submit a final, [43:12] that's uh that's good. I think that's [43:17] >> um on the chaos board level, don't we [43:20] need some approval to before we go on [43:22] our report? [43:24] >> I would imagine so, but just getting it [43:27] to the final PR stage is fine. um what [43:30] we would need now. Cali, do you have any [43:32] idea at this stage for just having a [43:34] repo? [43:36] >> Adding a repo underneath like that's [43:40] like technically under a specific [43:41] working group. I do not know. We're [43:45] through the process that I think Adrian [43:47] and I and others have been on. We're [43:49] starting to realize like, okay, how do [43:50] how should we actually like propose a [43:52] question to the board where they can [43:55] answer or respond to? So, we're hoping [43:56] maybe to get a better process of that [43:59] long term. I'm actually not on the [44:01] board. Um I don't want to [44:04] Sophia is, but I um so I actually don't [44:10] know what that is. Um whenever it's like [44:13] what is the difference between a [44:14] repository that implies a project like [44:17] an actual project or a repository that [44:19] is like a sub portion of the data [44:22] science working group. I think [44:23] personally I would want if the intention [44:26] was for this to like be a part of the [44:28] data science working group that that [44:29] would be integrated into the at least [44:32] the domain the title to be like this is [44:34] a sub pro like this is under the d like [44:37] we're working on this the data science [44:39] working group but it's under this [44:42] repository for xy andz reason I mean up [44:45] until like auger had like five different [44:47] like repositories and chaos's org and I [44:51] don't really know if there is a formal [44:53] process to requesting those or where of [44:55] what we should do because I have the [44:57] same question and the answer I'm not [45:00] surprised Sophia not from the [45:04] >> so great question um yeah I do feel like [45:09] >> that people just create them I don't [45:12] think that [45:14] >> yeah take that as how you want to that [45:16] is how it's has how it's been I think in [45:19] the long term as in like in the next [45:22] like 1 to 3 months because there's now [45:25] going to be like an infrastructure [45:26] committee that I think some of these [45:27] processes will be defined out. Um [45:33] oh um the the board is not like a [45:35] private thing. Um [45:40] >> I think it's just it's just a very loose [45:42] open source organization filled with [45:44] like half academics that struggle to [45:47] make the calls. So I think what we [45:50] should do in this case I had a similar [45:52] problem when I tried to set up the like [45:54] AIML working group and like for whatever [45:57] reason Linux Foundation had lost the [45:59] institutional memory of creating a [46:01] working group. Wild. So it took us way [46:04] longer than it should have. But in this [46:06] case what we should do is we have the [46:07] opportunity to state to the board that [46:10] we have for now a like R&D subreository [46:13] under this working group because [46:15] specifically it's going to report to [46:17] this working group. I think if it [46:19] becomes a higher domain repository that [46:21] implies there's a community and a [46:23] separate meeting but this is where that [46:24] will report to. So I think that's fine. [46:26] But when we propose that to them, I [46:28] think we should [46:29] >> within that say this is the steps that [46:31] we went through and [46:34] do it with the sponsorship in place. Um [46:37] and say like this is the higher level of [46:39] standard that they want to hold [46:40] ourselves to when putting a repo into [46:42] place. I would like that. [46:46] >> Yeah. Yeah. Sounds great. Um because [46:48] yeah, it's I would say that it's not [46:50] that the board is like some private [46:52] entity. It's actually more that it's [46:53] been a very it's very it's just has been [46:55] a small community and um because of [46:58] recent events where somebody has taken [47:00] to I think taken advantage of the setup [47:03] of the infrastructure within the [47:05] community and has been very um what is [47:09] the right word uh [47:12] retaliatory [47:13] um that a lot of things have had to make [47:15] have a lot more thought and decision-m [47:18] that has not been necessary whenever [47:21] them entirety of chaos up until this [47:23] point. It's usually been like a, oh, Sal [47:25] wants to make a repository for this use [47:27] case in the in the um working group. [47:29] Great, create the repository and then [47:30] the conversation ends. And I think that [47:34] it's chaos should mature. So there's [47:36] like that side of it. Um but also yeah, [47:39] I I but I can understand the perspective [47:41] of being like why is this board so [47:44] elusive? And it's [47:47] it's unfortunate that the current events [47:49] have painted it that way. I'll be [47:50] honest. [47:52] Yeah, but we can uh get a little bit [47:54] more information from them. [47:55] Theoretically, this group is a bunch of [47:57] maintainers. And if we're looking for [47:59] clarity on how we are being governed, [48:01] especially if you want to establish [48:02] repositories and projects, that's [48:05] appropriate. So, just reach out to them. [48:07] They're human beings. They're stressed [48:09] human beings, but they have emails and [48:12] they're on the Slack. So, [48:14] >> I promise they're not scary [48:17] again. And I'm not on the board. [48:20] Yeah. Yeah. S if you remember we started [48:23] thinking of putting Oscope in a better [48:26] structure because of uh while submitting [48:29] for funding especially for the SSI we we [48:32] identified that the way the rep was [48:35] currently structured is not going to [48:37] present us as a project that is ready. [48:41] So that was why we started saying okay [48:42] let's give it a name let's put it into a [48:45] structured environment. So we should [48:47] factor that in as we try to approach [48:50] okay we have a R&D project under science [48:54] working group you know [48:58] >> yeah so I think this I think in order to [49:01] be compliant to the LF we have to do it [49:03] in two phases so what I want us to like [49:05] we should have this meeting and like [49:07] make sure we've got a super polished R&D [49:09] version and in that we're immediately [49:12] reaching out to the board and saying [49:13] look we want to raise this repository [49:15] what would the steps be we're proposing [49:17] there's a sponsorship model behind it [49:20] and then we push it to the next level if [49:22] we can get there but uh yeah we won't [49:25] just go steal the domain we have to get [49:28] everything and this is my fault I did [49:30] this to this repository but everything [49:32] has to be like triple signed was my idea [49:35] and it's so irritating every time [49:38] >> like [49:40] that I created [49:42] >> good good [49:44] excellent [49:45] >> at the very least Um I think that there [49:47] will by the time you code through I [49:49] think that this process will be a lot [49:51] more defined out because I mean we're [49:53] having there is multiple different [49:55] things that are going on in chaos that [49:57] is requiring like more formal like how [49:59] do we create a repository what is the [50:01] domain like what is the trademarking and [50:03] and different processes that [50:06] realistically people have been in chaos [50:07] for a while thing. there's a lot more [50:10] people involved than it's still a small [50:13] community, but there's a lot more people [50:14] involved than there was two years ago. [50:16] And so, it used to just be like a couple [50:17] of people just being like, "Okay, I'm [50:19] just going to do this and I'm just going [50:20] to do that." And then things get bigger [50:22] and it's a a sign of like success and [50:26] and traction, but then you don't have [50:28] the process of like, okay, how do we [50:32] create a a new project? How do we [50:33] donate? How do we do these things? And [50:35] so it's [50:38] growing pains that I think are a sign of [50:41] the specific like like this where it's [50:43] like two years ago I would have been so [50:45] and I still am like oh my gosh somebody [50:47] else is wanting to define out a data [50:49] science project with a new data set that [50:50] I'm not involved with like directly at [50:53] all. That would have been like a [50:54] mind-blowing suggestion for me two years [50:56] ago. Um really like [51:00] so I like to look at it I don't know I [51:04] guess maybe more optimistically [51:05] positively than struggling through these [51:07] things. It's growth. [51:11] >> Yeah. Well it's cyclic. There's always [51:13] things like this going around somewhere [51:15] in some community and most of it is just [51:17] institutional memories. So we have a [51:20] really cool chance at the maintainer [51:22] class right now to have some like [51:24] communication in making this process and [51:26] we can define what looks and feels good [51:29] to us, what processes we want to put in [51:31] place and I think the governing board is [51:34] going to be nothing but grateful for [51:35] some guidance. So [51:37] >> speaking of we have this call turns into [51:40] the community call in four minutes and [51:42] so if people are around I actually don't [51:44] think this is a bad thing to bring up [51:46] and the community call of being like [51:47] okay how do we formally go about [51:49] requesting like an addition to a [51:52] repository. Um while I don't know if [51:54] we're going to get like the formal [51:55] answer then I think that's a good thing [51:57] to like keep down outside of even like [51:59] the six people in this group. um as I [52:03] don't know if people have saw the [52:04] announcement on now we're having an [52:06] infrastructure committee um and so [52:08] that's pretty seems pretty heavily in [52:11] the domain of what what does the [52:13] infrastructure how does the [52:14] infrastructure committee want to handle [52:15] something like this and slash the board [52:17] and figuring out an easier process so it [52:20] doesn't have to be so hands-on for [52:21] everyone [52:22] >> I was just going to recommend that so [52:24] yeah like I think like we [52:27] >> in three minutes [52:28] >> in three minutes it's mostly like I [52:30] think that [52:32] committee has been proposed but I think [52:34] the actual tasks and jurisdiction is a a [52:37] little bit yet to be defined. So I think [52:40] asking right now would this be under the [52:42] purview and in scope for this committee [52:45] is like I think that's just a direct [52:46] question we can just ask because it's [52:48] going to be brought up on the next [52:49] agenda. So we might as well get clarity [52:51] on that. [52:51] >> Can I be incredibly lazy and ask someone [52:53] to throw a link to that next call in [52:55] this chat? [52:56] >> It is here. You don't have to move. [52:58] >> Oh god. Oh yeah. I that's my least [53:01] favorite thing about Chaos. [53:02] >> Really? [53:04] >> Yeah. [53:05] >> Why? [53:06] >> I mean, it's the living room style and [53:08] it's the way that Godfly works, but like [53:10] I don't like that there's like meetings [53:11] in the living room style. I just want [53:13] that to be Oh, whoops. I popped in. [53:17] >> So, but uh Okay. So, [53:19] >> so this this will just transition and [53:21] it's also the same exact document. You [53:23] just go to the community meeting portion [53:26] of the document. That's true. and and [53:28] then the ag you can just add your agenda [53:30] point in there. [53:32] >> Um [53:32] >> it's not a bad process. Okay. Well, in [53:34] that case, I will jump on for 15 minutes [53:36] of the next one and then I shall end my [53:38] day having made my case. But I must run [53:41] to the restroom first. I will be back. [53:43] >> Okay. I am facilitating so I will put [53:47] that first. [53:49] >> Oh, I'm [53:52] Callie and it's the show. [53:56] Hello.  📄 **Full text:** ![[Full Document Contents/Articles/CHAOSS Data Science Working Group, April 21, 2026.md|CHAOSS Data Science Working Group, April 21, 2026 by YouTube]] ## Highlights - this. We've already had these com I'm like, "We've already been through this." I agree. Um, we have a team that's looking at that more closely. Um, I have like consultations with them. Um, I wish they would publish their findings more. I might poke them on that. um because >> this would be a really great place to do that and would be very happy to collaborate on that type of I mean that's what I'm hoping to be able to do um as things move forward hopefully with ([View Highlight](https://read.readwise.io/read/01kq2r03psjnm9gr2tkmgsv2dk)) - Tags: 👻 ai highlighted - then I've done all the work already. Like I went through and like added 200 labels to open-ended questions and then it was like okay now I could summarize it and it's like yeah I did all the work. >> Yeah, I could have just done this myself. Yeah, I think that's as like a whole I'm going to be very curious to see where like data science and more like um what like and pretty much just like strategic analysis like an and where you're actually using an you're using ([View Highlight](https://read.readwise.io/read/01kq2r03q0e9ssrg5219s82qmb)) - Tags: 👻 ai highlighted - attract more collaborators. So one thing we notice is the issue we opened on the data science working group is attracting contributors but most of those contributors they don't come back. Yeah. So those are the those are the stages that we are I'm currently thinking of how to you know bring them coming back you know I'm going to discuss that maybe later today then we could probably share opinions on how that could be solved and all. So that's pretty much it and um we are happy back. ([View Highlight](https://read.readwise.io/read/01kq2r03r72tgf00kn607t9jw0)) - Tags: 👻 ai highlighted - that is like kind of bringing all the sources for this project together and bringing it under like an like a a a name. Um and so it might be a good point to like where everything is at, what the definition of the project is because I think that's um to kind of formalize it a bit more. And I think there's there is like a kind of a more broad discussion I guess for I'm I'm really liking how the data science working group is like shaping up to be but like okay there's ([View Highlight](https://read.readwise.io/read/01kq2r03stc5bh9vnwrbrn3xga)) - Tags: 👻 ai highlighted - going to be constantly like like sub projects under this coming in and out. How should we structure this? How like how do we structure the repository so it makes sense to people to go into different the different areas? Um which I'm I said I'm really happy that this is how it's turning out to be. Um but how do we how do we handle that? Um cuz I can see that there is now like a new data set that is around that's really interesting around like corporate um investment or involvement in open source ([View Highlight](https://read.readwise.io/read/01kq2r03ts9eskddjx6ec87nfb)) - Tags: 👻 ai highlighted - sample data in it for people to play around with. >> Thanks for sharing that. I'm ready to test it when you're when you're there. Yeah, I was the one that added the question of how do you retain contributors? Although I'm not I'm not talking broadly. I'm not talking based on the the issues that I think we are ([View Highlight](https://read.readwise.io/read/01kq2r03ygkyq51rcyf2tnfegx)) - Tags: 👻 ai highlighted - think the efforts around how do we attract and retain contributors? It's how do we have meaningful work for people to be able to do because a lot of times we do have a lot of people who comment on um in the data science working group in different context of like looking for um yeah looking to do data science work. And so I guess my personal perspective is like instead of we we only have like a very small set of people instead of trying to put time towards like a ([View Highlight](https://read.readwise.io/read/01kq2r043px5pyh4gj8xwwfcdq)) - Tags: 👻 ai highlighted - >> got to be careful. >> Yeah. um that it kind of ties in well to what we were some of the stuff that we were talking about um before you had hopped on the call of how do we like we're now starting to gather some data sources. How do we make it to where there are consistent ones that people can use for different use cases? So this all fits in pretty well with what um we were talking about before cuz it is very interesting that there the problems that we have in data science are still ([View Highlight](https://read.readwise.io/read/01kq2r047tjhhsqwdjr8yz7kyd)) - Tags: 👻 ai highlighted - to I think taken advantage of the setup of the infrastructure within the community and has been very um what is the right word uh retaliatory um that a lot of things have had to make have a lot more thought and decision-m that has not been necessary whenever them entirety of chaos up until this point. It's usually been like a, oh, Sal wants to make a repository for this use case in the in the um working group. Great, create the repository and then ([View Highlight](https://read.readwise.io/read/01kq2r04mm3ev9jcct77y4y4z6)) - Tags: 👻 ai highlighted