Windows NT will be with us longer than systemd and flatpak.
For me, walking through an old Souq gives me a similar feeling of awe / mortality / insignificance as viewing a cathedral or looking from the Colorado ranch land up to the Rocky Mountains.
Also some cathedrals have remained "Catholic" since their raising, but there are a lot that have changed from Christian to Islamic to Protestant ... both the cathedral and the bazaar's physical buildings are still present from the same era and both are used for their original purpose (marketplace or worship). And both have delibly shaped their regions by being engines of culture, innovation, and power.
However ... the domain of operating systems is subject to weird constraints, and so it's not really appropriate to make some of the observations one might make in other domains. Nevertheless, I thought the point was that we want things to improve via replacement (a "bazaar" model), rather than stand for all time. We don't actually want technology "cathedrals" at all, even if we do appreciate architectural ones.
Are you referring to the *BSDs? Linux isn't Unix derived. It's a re-implementation with no code traceable back to the original Unix or any of its descendants.
Aside from that, Windows has a revenue stream behind it that looks to continue indefinitely paying for its development and Linux has...? The half-digested carcass of Red Hat within IBM? Canonical?
Linux is the root OS of Android, which has a lot more revenue behind it than Red Hat or Canonical.
It's no longer a bazzar where random university lecturers write and run entire subsystems between classes, but it is at least multiple companies contributing.
This type of informal market likely outnumbers cathedrals, especially if you count the ones that evolved into tourist markets, high streets, malls, and central business districts.
How many bazaar projects from even 10 years ago are still maintained? Go through GitHub's trending repos from 2015. Most are abandoned. The successes transform - GitLab, Linux, Kubernetes, more Cathedral than Bazaar.
Did BSD defeat Linux? No. Which BSD is even the right one? BSD's biggest success is living on as the foundation of Apple's Cathedral in XNU, and PlayStation's Cathedral in the PS4 and PS5.
Did Linux stay a bazaar vendor? No - 90% of code has been corporate contributed since 2004. Less than 3% of the Linux Foundation budget goes towards kernel development. Linux is a Cathedral, by every definition, and only exists today because Cathedrals invest in it for collective benefit. It's a Cathedral, run as a Cathedral joint venture, to be abandoned if a better thing for the investing Cathedrals ever came along.
GCC? Being clobbered by Clang. Less relevant every year. Same with GNU coreutils, slowly getting killed by uutils.
Firefox? Firefox only still exists because a Cathedral called Google funds it.
LibreOffice, Apache, PHP, Blender? Professional foundations that get very picky about who is allowed to contribute what. They aren't amateurs and they all depend on Cathedral funding. Blender only got good when it started collecting checks from Qualcomm, NVIDIA, AMD, Intel, and Adobe. Blender is a Cathedral funded by Cathedrals.
I really dislike all the corporate involvement in Linux. I don't believe in win-win with commercial. That was the main reason for my choice though there's other things I like too such as full ZFS support and great documentation.
Uhh, all the big ones in common use? GNU’s massive portfolio of software, Linux, multiple BSDs, Apache, Firefox, BusyBox, PHP, Perl, the many lineages of StarOffice, LaTeX, Debian, vim, fish, tmux, I mean this barely scratches the surface. Are you kidding me?
How many startups have failed over the last decade? I would argue that the norm is for any project to eventually cease. Only useful things with an active community (whether that community is for-profit or not) tend to last, until they are no longer valued enough to maintain. This goes for things in the physical world just as it does for software.
I'm not there yet, but I am perilously close to tipping over into believing that making open source software today is actually doing harm by giving more free labor to an exploitative ecosystem. Instead you should charge for your software and try to build an ecosystem where the customer is the customer and not the product.
I stress today because this was not true pre-SaaS or pre-mobile. FOSS was indeed liberating in the PC and early web eras.
I recently had a realization: I can name Cathedrals, that are 800 years old, and still standing. I can't name a single Bazaar stall more than 50 years old around any Cathedral that's still standing. The Cathedral's builders no doubt bought countless stone and food from the Bazaar, making the Bazaar very useful for building Cathedrals with, but the Bazaar was historically ephemeral.
The very title of the essay predicts failure. The very metaphor for the philosophy was broken from the start. Or, in a twisted accidentally correct way, it was the perfect metaphor for how open-source ends up as Cathedral supplies.