Sprints/Goals/2024/Eco Notes

Kick-Off of Project "Sustainable Software for Sustainable Hardware"

Presentation

Slides

Notes

• what would be use cases for very old hardware?
• did you talk to to distributions already?
• focus on specific use cases? specific distributions?
• releases have to be done regularly.
• is steam os good for normal consumer?
• which is a good distribution for consumers?
• does kde run on old kindles? ebookreader, are they possible to reuse...as picture frames etc.
  • talk to hardware vendor about this problem.
• statistics about which are the most unused devices because of lock down.
• this is our minimum hardware and this is how old it is.
• how far do we go back on hardware?

• knock on effects. if you take sufficient software from an old hardware onto an new hardware, the battery will last a lot longer (plus point for consumer)
• what is the difference between old hardware and new hardware. (ecolab)

• Documentation for developers on what works on what computer
• find more companies that would support the idea, repair companies, refurb companies (workshops etc)

• what documentation is out there already?

Statement: your computer is broken, we can repair it.

Meeting people where they are:

• How do they use their computer day to day?
• What HW do they have now?
• What can the envision for using hardware for? Use cases?
• What may hold them back? Trust? Technical skills?
• What handholding would they want/expect?
• What terminology makes sense to them? Vendor dependency? FOSS?
  • Right to repair seems to resonate, what else
  • Sustainable software
  • Is there data? Finding studies to look at this
  • Umwelt Festival survey, also qualitative conversations (questions to guide conversations), feedback on leaflet, write down their questions write a FAQ based on this (do this systematically)
  • -> make a pad/invent issue to structure the input, put a concrete goal, establish criteria, avoid leading questions
  • https:://survey.kde.org for doing a questionnaire for the survey
• GOAL: we want them to install Free Software, they want to keep using old HW
  • Focus on desktops
  • Differentiate between installing and using HW
  • What HW do you have now? What software do you use now?
  • Demographic info: Age? Gender? Computer literacy?
  • What do you do with your computer? Name applications that are very important to them:
  • What is important to people? Rank in terms of importance: Battery life? E-waste? Ads? Tracking? Security? Saving money?
  • Do you have old computers/laptops around?
  • Completing survey can give them something to do
• Applications to install? LibreOffice? Okular? Alternatives to proprietary software
  • What proprietary software is installable
  • See also https://prism-break.org/en/
• Steps to dabble
• Values users have that overlap with FOSS
  • Independence, Sovereignty, Sustainable
  • Onboard people to often used terms (how can the term FOSS resonate)
• Related projects
  • Zorin OS (https://zorin.com/) (they mentioned reusing old computers, but they are a GNOME distribution, not KDE)
• What would stop you from using FOSS applications? Operating systems?
  • What would you miss if you gave up your computer today?
  • Installation
  • New OS patterns / interface, takes time to get used to
  • Ranking the severity (don't know how to install an OS, security issues)
  • Observe users as they use it
  • What will happen to my data?
  • Dependence on proprietary software
  • More likely to try it on an old device, not replacing the current one
  • Not understanding enough about ccmputers to feel fully comfortable
  • Lack of support
  • Lack of knowledge about software options
  • Don't know anyone who uses the software, no one to ask for help
  • Data "stuck" on a device
• Installation video?
• Places for workshops:
  • repair cafes
• Get inspired by tradition/history/experience of Linux Installfests / Installation Parties
• Groups of people
  • go to where they are
  • bring them to hacker/makerspaces
  • multipliers (repair cafes, teachers, public libraries...)
  • https://linux-hardware.org/

Berlin also has several useful public libraries, many of which provide computer access. These may be good places to also do outreach. Even leaving a computer with FOSS running on it would be a good way to showcase the capabilities that can be accomplished on a budget.

• Blog post publication channels
  • Goals Sprint should be prominent
  • New project funding also

Questionnaire

We want data on attitudes to reuse of computers. For that purpose we discussed how to make a questionnaire that can give us useful information. We don't want the questions to be too technical, more to give a broad overview. Questions to consider includes:

INTRO: WE ARE INTERESTED IN .... more sustainable ways to use hardware

• What is your current computer?
• How old is it?
• What did you use before?
• Why did you switch? (buy a new one?)
• What would motivate you to switch again? Beautiful design, new hardware, ...?
• Were there any obstacles for switching in the past?
• What would be the obstacles for switching the next time?
• Would you like to keep using your current computer?
• How willing would you be to try a new system?
• How much effort are you willing to put into keep using the computer?
• How willing are you to change?
• Would you use new software that does not look or work quite the same way as you are used to?
• How much time do you spend on your devices? (Laptops, Phones, Tablets, Gaming Console)
• What is important for you when deciding for a new computer? (Newest computer? Sustainability issues? Beautiful design? Nominal scale 1-10)
• Do you manage your own computer?
• Do you have support?
• Some profile questions

We want to know what we need to provide to make it attractive for people to install a free OS instead of buying a new computer when planned obsolescence sets in. The questionnaire will be tried out ...(Joseph, please fill in details)...

[2024-04-24] Draft of questions at Invent: https://invent.kde.org/teams/eco/ns4nh/-/issues/5#note_928617

Purpose

The largest environmental impact from computer use is production of the devices. Proprietary systems tend to require users to buy new hardware often, even though the hardware is perfectly usable. Therefore the greatest impact we can have in this respect is to provide software, that people feel comfortable to switch to. There is a technical aspect to this, but also a social/psychological one. We try to find out how to frame our message.

We might be able to find new allies, who have not traditionally thought of computers and free software as something relevant to them, like environmental activists, community groups, schools and libraries.

Ideas

Do a competition in the community who can get KDE to run on the oldest hardware. Will help to spread the word in the KDE community, give us data about on which systems KDE runs, might help with improving install instructions. And could be a lot of fun.

Graphics cards consume a lot of energy, even when unused drivers still run in the background. We can't do anything about it because drivers are proprietary. Older graphics cards cause bugs: after a while they are no longer supported by the producer - instead they advice costumers to buy a new card. Could we try to raise the issue politically?

Website

https://eco.kde.org

• Developers (join the community, eco-certifying your software) / End users (awareness raising and installation)
• Why should you care about this?
  • Reason why and what are we at KDE doing
  • Pyramid structure: Hook, Details, Call to action
• Blurry lines between target audiences?
  • If we want to attract 'normal' users, we should be user oriented
• We did a collection of sketches for the site. Will continue on Sunday to consolidate and come up with a wireframe for the next iteration of the site (somewhere are photos)
• We did a wireframe for the next iteration of the eco.kde.org website (somewhere there is a photo of the whiteboard)

Sustainable AI

Presentation of thoughts about adding "Ethical AI Rating" to create transparency about the use of AI: https://invent.kde.org/teams/eco/sustainable-software-goal/-/blob/master/presentations/2024-04-21-sustainable-ai.odp

Notes from the discussion:

Would we even want application less than green in KDE? This probably is an additional discussion, because what to include or endorse from KDE's side will be informed by how AI is used and there are open questions. But it still would make sense to make things transparent, also for applications outside of KDE which are based on our franeworks.

Is a model, which is green, actually always in line with the spirit of the manifesto, and a model which is red not? A model trained on proprietary data which generates material to be freely used, which would be red according to criterion 3, could very well be in our interest.

What is a derivative work in the context of models trained on GPL-licensed content? Is this a way to circumvent the GPL? What should our stance be there? Would we want it to be a derivative work? Legal questions are still kind of unanswered.

What are the implications in terms of resource-efficiency? Would AI features prevent our software to run on old hardware? Would we want to advertise AI features if they need for example modern GPUs? It depends a lot on how useful these features are. There are differeing opinions on the usefulness. For some cases such as translations we agree that it's useful, but for others we don't.

Important but separate discussions: Do we want to use AI for our internal needs, such as translating our applications. Is this ethical? How do we want to publish/train/license models we create ourself.

We should broaden the discussion within KDE but also beyond that. For example Krita is thinking about AI features. Or Nextcloud and Mozilla. Motivations might vary. KDE actually is in a position to not be driven by the hype, but a balanced view on what actually makes sense in which situation.

More Material:

Mozilla's stance: https://assets.mofoprod.net/network/documents/Mozilla_Accelerating_Progress_Toward_Trustworthy_AI_v3.pdf Another labeling approach: https://nutrition-facts.ai/

More Notes From Sustainable AI (unedited)

Q(uestions) & A(nswers) & C(omments)

• Q: What do you mean that we should adopt the ethical AI rating. A: For AI in KDE software we should show this scheme in our software.
• Q: I agree that the ethical AI rating makes sense in the context of Nextcloud. But is anything below Green compatible with the KDE manifesto. Red is comparable to supporting GMail. But yellow and orange are in violation: we are bundling non-free code. Green is the only thing we can include with our software.
  • C: Same issue around training data.
  • Q: Do we assume output is a derivative work of the input? A: We cannot answer that. Q: Ok do we want the output to be a derivative work? C: GPL allows for payment by contributing to the code, but this is not possible with AI.
  • Q: What do we want here? We want to respect other people's licenses.
  • C: For copywrite, the issue is expression (output?)
• C: Once we let the genie out of the bottle, we may not be able to put it back in.
• C: Let's go over the criteria again. By making this information available, we let end users make decisions.
• Q: If a ML model uses our code, should output be licensed under GPL?
• C: What stance do we want to take? Does AI copy from us or not?
  • These are philosophical questions.
• C: Is using our code something that we can restrict?
• C: (in chat) In addition to use of AI in KDE apps, possible other areas of direct concern:
  • Use of AI for translation, what can be incorporated, guidelines would be helpful
  • Use of data associated with KDE projects in creating AI models. If KDE does this what conditions should be associated with it, this could be useful in improving energy efficiency and app effectiveness -> Questions to address at a future date.
• Q: How do we mark the output so users are aware that this is AI generated? A: This is another dimension that can be addressed in concrete examples.
• C: We should make sure our conversations are aligned with other discussions, like with Krita.
• C: Let's find the way we can integrate AI so that we can support this, let's not be carried away with the wave.
• C: I have no problem marking things as red or orange, it is the green marking that I am uncomfortable with. If input is CC0, output is CC0. That is clear. But if input is GPL, should the output be GPL.
• C: These conversations are certainly happening elsewehere, we should join conversations.
• Q: What are the eco considerations here? Do we become dependent on AI features and obsolete old HW? A: These are the same general questions for other features in KDE.
• C: We can make decisions like Plasma by default doesn't require these features. They are add-ons, not core features.
  • C: As long as there is interest in supporting older devices, this will not be a problem.
  • C: Issue of using AI as a solution for everything. This can result in dubious value-added features.
  • C: Policies / guidelines: If this can be done without AI, do it without. C: This is not needed. It will happen when reviewing the code. C: But guidelines will prevent a fight about this.
• C: For marketing, focus on efficiency gains in training AI (e.g., so it can run on rasbpi). C: But only if this solves an actual problem. For instance, translation or transcribing.
• Q: Do you project into the future, do you not think AI mdoels will get better? A: My answer is it is not economical.
• C: KRunner lists can be informed by an AI. That would remove buggy heuristics. We can tweak on top of it.

REMOVING BARRIERS, WHAT CAN PEOPLE DO