Kiwix's 20th anniversary legacy

[rgaudin]

Around this time 20 years ago, @kelson42 and I were probably putting together the first pieces of what would become Kiwix.
We were good people, doing good things with good intentions.

20 years later, I'd like know what is Kiwix's mission, its strategy and whether our decisions and behavior are considered aligned with it.
I am frequently wondering what those are and am often convinced we're not aligned with what those could be.
Clarity on this would help solve cognitive dissonance for those who care.

Here are some clues at why I disagree with “same as has always been and are perfectly aligned”:

• We are obsessed with sales
• We are stating sale objectives that are unrealistic.
• While our tools' objective is to empower users by helping them access information (mostly via text), we consider our visitors don't diserve to be treated as capable of reading text.
• We are designing our websites so that a purchase is the first thing our visitor are presented with.
• We always regard competitors with disdain
• We use Gen AI
• We don't meet anymore
• Decision making has shifted from horizontal to top-bottom
• We are isolated in silos, unaware of what others are doing
• We don't know general priorities anymore
• Roles are unclear or obsolete
• We don't know what's in the pipe anymore
• We consider taxes on labor a bad thing
• We pride ourselves as being a non-profit that works like a startup
• We have ceased to mention environmental impact
• We are not happy that people are integrating our tools because they make money off of it
• We are discussing dual licensing
• We are saying “b2b is priority” but only focus on b2c
• We let Google track our visitors for months
• We store our org data at Google and our code at Microsoft
• Our main target is Preppers
• We're seeing that non-trivial topics are left unanswered
• We are investing more energy in selling our products than in improving them
• We see erratic change of position to some topics
• We're seeing more messages that are inconsiderate

Non exhaustive list but should help picture my biais.

---

Note that this ticket is intentionally in the Open and is addressed primarily at @kelson42.
It's not a trick question. We all attach unspoken meaning, values and conditions to the stuff we do, especially when it's not been clearly defined.

I would like to reconcile my personnal opinion with the current trajectory and for that I need to know if I need to make peace with it or if the steer has been left unattended.

[kelson42]

@rgaudin Thank you for your message.

Already a small feedback:

• Agree with most of the points
• Overall core situation has indeed changed (in comparison to pre-2015 situation) and we have to embrace it (not live in the past)
• Always interested in alternative approaches
• Many of the points bound to my lack of time/investment
• Been nice is essential, but facing the reality and not being stupid as well.

I propose now to have a live discussion and transform these points on actionable when possible.

[kelson42]

We let Google track our visitors for months

AFAIK this has been fixed!? See kiwix/web#147 and kiwix/web#99

[benoit74]

Thank you @rgaudin for bringing this to the table. I feel the friction on all points you've raised. I'm sure I'm not aligned with you on what Kiwix should be on few of these points, but this is part of the friction I feel.

I do not intend to provide much personal feedback on this for now unless you request it, but I can't refrain from providing personal reactions to first answers.

Many of the points bound to my lack of time/investment

I disagree. Kiwix can and should adapt to this, we have way sufficient funds and capable persons on board, and we can even find more. IMHO what Kiwix misses is that it did not adapted to this, at least not in the correct way. I remember having raised the concern in 2023 when I joined. With proper focus limited time/investment is not a concern.

Not having the same investment possible after 20 years is expected. Not adapting is a concern.

I propose now to have a live discussion and transform these points on actionable when possible.

I'm interesting in at least the outputs of this discussion. I'm available to help if needed (and if I can).

I consider this is also something to discuss with the board (some of these concerns should even have been raised by the board).

Please focus on high level outcomes first: for instance I don't mind we turn off data sent to Google now, this is a "detail", I prefer we first change Kiwix so that either it will not happen again or we recognize it is necessary.

Start with why - Simon Sinek

[Jaifroid]

I'm a bit of an outsider, being a volunteer and in full-time employment in education, so I don't feel it's my place to comment on the organizational ethos so long as it still has a place for people like me. I've certainly seen a shift since I started contributing to Kiwix JS around 9 years ago now from what felt like a volunteer-driven community (I may have had a distorted view, given that the person I was working most closely with was also a volunteer) to one more focused on raising the revenue needed to keep the servers running, the software patched/updated and to pay people's salaries. But I also recognize that the sheer amount of data users are downloading and the cost of generating ZIMs on the massive scale you've achieved is not something that could have been maintained without a significant income stream.

The only thing I'll comment on for now is the stance on AI, because I don't think that genie is going back in the bottle. It can be massively useful (no way could I have added new C++ bindings to javascript-libzim if Claude hadn't helped me figure them out by investigating the Node JS and Python equivalents), and it can also be a massive pain (I've seen too many mindless AI-generated PRs that contributors don't even bother testing, over at Kiwix JS). This is no different from any new technology: it has disruptive, even destructive, effects alongside its considerable benefits. I know this more than anyone seeing how it is transforming students' approach to their own learning and writing in Cambridge, and how it is disrupting our assessment and forcing us to revert to old-fashioned "technologies" such as writing exam essays by hand.

One thing I know for sure is that we can't ignore it, given the huge changes going on around us. People are now vibe-coding their own Kiwix clients in a few hours for unsupported hardware such as e-tablets. Is that a bad thing? Or does it democratize the generation of code, even if the vibe coder doesn't understand a single line? These questions go way beyond Kiwix, of course. My personal approach has been to try to "learn" AI: understand it intellectually and practically. I believe that you're only a slave to something you do not comprehend and whose power you do not know how to leverage to good ends.

And the final thing, having known some of you for quite a long time now: you're all good people, and behind any differences regarding the means to achieve its ends, Kiwix still appears to me to be working towards the same overall goals that it was when I joined 9 years ago. It is also good (and important) to have a courteous debate on the desired means to achieve those ends.

[Popolechien]

I'll add my own reflections on this, but would reframe the question as What is it that is that we did right and Where do we still fall short?

We are still true to our mission: free knowledge for people without internet access

• We are not “just” the offline Wikipedia people anymore
• The use cases have greatly expanded - from Malian schools, Kiwix is now deployed in schools and libraries across the globe, smuggled into North Korea and downloaded in China, Iran or Russia; it runs in Antarctic bases, prisons, and military-grade air-gapped facilities.
• All code produced is still FOSS.

Kiwix and openZIM have become standards and the tech is reliable

• Gone are the days where the decision to use Kiwix or XOWA or Wikitaxi or any other offline reader was not self-evident.
• We are now in a place where the ZIM format has become or is becoming an accepted standard for offline storage of working websites. Orgs create their own ZIM files without asking us about it. People use the ZIM format for offline RAG training while we’ve never really looked into it.
• All platforms run as intended
• Businesses are being built and run on the back of Kiwix
• Zimit is already a quantum leap that is yet to deliver its full potential.

We have good, professional governance

• Kiwix has moved from “friends writing code when they can” to “people’s salaries depend on it”.
• Our accounting is clean, we can produce any official document upon request, and the association is not working under the assumption that we better not get asked difficult questions
• People are comfortable saying things that are not pleasant to hear, and are willing to hear things they don’t agree with/don’t like.

Our core community is still small

• On the plus side, we’re close-knit and there’s not enough people to have drama.
• But this also means that most repos are maintained by a single developer/single point of failure.

Our funding model is fundamentally weak

• When 10 salaries are on the line, money becomes a central issue, like it or not.
• 50-70% of our revenue still comes from the Wikimedia Foundation. The amount of juggling needed because they got 11 months late in processing their 2023 payment is not something I want to live ever again. Yet, we have few other recurring donors.
• We lack a stable of clearly identified projects that could both be sold to potential funders and help us grow.
• Individual donations are abysmally low.
• We do not have a clear understanding of what our moat is.

Users and user experience are still an afterthought

• While extraordinary efforts are put in place to document our code and processes for fellow developers, very little is done to accommodate low-effort, low-proficiency end-users with no mastery of English.
• We develop products and features based on feels and keep quite a few of them for historical reasons. We fully disregard any little data we have if it does not meet our hobbyist approach. Kiwix has no such thing as a high-profile failed and shut down project.
• Content discovery is inexistent.
• We still do not know our users. We do not count them, we do not collect feedback in a manner that helps drive development.

There is no long-term strategy, nor leadership by intent

• Everything is treated as a priority, new projects are continuously added to the pile
• We do not have a shared framework that allows people to make independent, informed decisions that will help us scale.
• Each tool or platform we work on or maintain should have a stated purpose, reasons for this objective to exist, and a clear list of constraints and resources available so that everyone involved can then independently try and figure out how to reach their objectives.