For one thing Photoshop is a transferrable skill. Gimp will get you laughed out the door of most photo studios and design agencies.

For another - too many FOSS people just do not understand that tinkering with tools is not the point for most users.

If you like tinkering with tools, go tinker with tools. Most people have work to do, and tinkering with tools is not something they want to be distracted by.

Photoshop is far from being my favourite product, and Adobe are very far from being my favourite company. In fact I - and millions of other users - would be delighted if we could save $$$$ a year with a free alternative.

Which applies in the general case. If FOSS was a realistic and workable alternative, it would storm the market and most people would be using it.

But it isn't. Most FOSS applications are aimed at self-selecting technical users, and there's very limited interest in making tools for non-technical users.

In fact IMO FOSS culture lacks the mindset and skills - and very possibly the empathy - to understand what general users want from their tools.

It's been a success for developers who want to make technical tools for other developers. And also for infrastructure. Not to take anything from that, because that's both significant and true.

But for popular non-expert applications? Simply - no.

Yeah, exactly this. Open Source developers don’t care about how you work or what you want your tools to do. They care about what’s fun and interesting to program.

I can’t imagine forcing students to care about .rc files and whatever’s in /etc. Not to mention the differences between whatever shell is installed by default and the various other shells that are also scripting languages, etc. I’m a veteran programmer in my 50s and this shit is too much for me to care about. I have work to do. Editing text files in vim or Emacs is utterly stupid for most users. It’s not like TextEdit or Notepad don’t do most of what students need without a bunch of cryptic crap involved, or that IDEs wouldn’t allow students to delve into programming much more easily. I just don’t get the cult of Unix. Yes, it’s powerful for a large class of things that programmers like to do. But it’s a terrible OS for normal users. And that won’t change until the people making it and making tools for it grow up and think of someone other than themselves when they write programs.

While I agree with this, it's interesting how Blender basically resisted losing focus on boring-but-important things (while keeping the fun and interesting parts). Probably the fact that Blender was originally planned to be a commercial application.

If you could get krita/gimp/whatever "close enough" to market for industry to "see the potential" then maybe similar could happen.

Although, for all I know, there is a dark side to the Blender current ways of doing things. But for now it seems fantastic.

It also helped 3DCG is relatively niche, at least not the type of a tool that people would expect to be able to use it without learning it considerably. At the same time Blender did make some compromises around the 2.8 version to adopt to more of industry standard UI interactions as opposed to sticking with the their own way of doing

> Yeah, exactly this. Open Source developers don’t care about how you work or what you want your tools to do. They care about what’s fun and interesting to program.

It's more likely that designing effective UX is just extremely difficult. I'm guessing it's so difficult that neither the user nor the FOSS developer even know how to define the problem in a lot of cases.

E.g., go back and try to find a pre-Google era user asking for streamlined search capability in some Linux application mailing list. Something like, "I just want to type in the thing I'm looking for and find the thing I wanted." IIRC most university library search tools before Google had a few dozen fields to specify the type of data you wanted to search for, and a crappy "keywords" field that often didn't return the desired record. So a student would search by first not finding what they were after, then calling over a librarian to help them choose the correct fields and input strings.

I'll rankly speculate that any such discussion on a pre-Google FOSS application user list would have consisted of a developer arguing that such an approach cannot work by providing a set of canned edge-cases with unresolvable ambiguities. And this response would have convinced the user who likely had no idea how the whole search algo works in the first place, or what the potential solutions could look like.

Then Google came along. (Well, the Google that exist before the current one which seems to be purposely inserting non-relevant results that a recommendation algo has determined still have a decent chance of engaging the user.) And the FOSS mailing list response tacitly shifted from, "That's probably not a well-formed feature," to, "Patches accepted." I'm not sure what the current response would be, but I doubt it generates a lot of enthusiasm for improving the UX.

Hell, even in the well-formed feature requests the implementation details are probably a major pain in the ass. For funsies-- go download the old Gimp that consisted entirely of toplevel windows, and try to figure out the most workable way of converting it a single-window app with subwindows. That's like some kind of demented detention punishment for FOSS devs who break the CoC. I wouldn't wish it on anybody.

This is pretty much a self-perpetuating cycle and what happens when you have something close to a monopoly.

Everyone uses Photoshop, almost nobody uses GIMP. Everyone uses Chrome, almost nobody uses Firefox. Everyone uses Windows, almost nobody uses Linux and so on...

Thankfully, in the server space, *nix reigns supreme and FOSS has a larger foothold, though thanks to the above, the "year of the Linux desktop" may as well never come.

> Most FOSS applications are aimed at self-selecting technical users, and there's very limited interest in making tools for non-technical users. In fact IMO FOSS culture lacks the mindset and skills - and very possibly the empathy - to understand what general users want from their tools.

I'd say that it's also a matter of prioritizing what's necessary to get things done and simply not having the resources for the rest: UX, marketing, branding, integrations and so on.

That's simply not true. While having high market share does give you a short-term advantage, if a better alternative exists you'll bleed market share over time. Look at what happened to Netscape, then Internet Explorer, then Firefox. It'll happen to Chromium eventually too.

Open-source alternatives to popular desktop software packages have existed for decades, and the reason they haven't snatched market share away from the main packages is because they don't work well.

I remember being surprised when a couple of older people I was building computers for asked me to put Office on them. There was only one guy I knew from my generation who still used it.

The time frames at play shouldn't be forgotten about here, nor should the magnitude of the effect.

Most developers universally hated IE from IE 6 (2001) to IE 11, which might still be supported until 2030:

> Microsoft is committed to support Internet Explorer that way to 2030 at least, with one year's notice before it is discontinued.

https://en.wikipedia.org/wiki/Internet_Explorer_11

And yet, it still had to be supported and for some systems still needs to be. For whatever stupid reason, many were locked into using it and had little choice. Sure, eventually it could be tossed aside by most sane enterprises, but if developers had been given proper reign over what to choose to use, then it would have been ditched a few years after the release of IE 6 (Firefox came out around 2002).

The average Joe/Jane doesn't really care about the browsers, though. They might not even recognize what a browser really is, merely know that they can use that one icon to access the Internet. Thus, the choice is made for them. Similarly, most laptops and desktop computers come with Windows preinstalled due to manufacturer deals, much like how most Android variations come with adware and other garbage. Nobody wants it, but nobody (at a certain price point), has any alternatives.

If you don't know that there's a choice to make, someone else will make it for you.

And i'd posit that there isn't as much of a difference between being technologically illiterate in Windows and being technologically illiterate in *nix.

> Open-source alternatives to popular desktop software packages have existed for decades, and the reason they haven't snatched market share away from the main packages is because they don't work well.

This is fair, though, especially because most FOSS software is basically developed below the poverty line: https://staltz.com/software-below-the-poverty-line.html

Nobody's going to have good UX or fancy onboarding if they can barely find enough resources to keep the project going and fix all of the bugs, as well as handle the technological churn of keeping dependencies up to date etc.

Outside of the ravings of FOSS enthusiasts and niche circumstances (e.g. servers), open source software will generally find it hard to compete with commercial offerings, most of the time, especially when it comes to software for the common folk.

I hear this a lot, but I don't think it's the primary reason why Desktop Linux has negligible market share (neither does Linus, FWIW). Plenty of people build their own PCs, or have a custom PC built for them by a friend/independent shop, and even amongst these enthusiasts Linux use is rare.

The main problem, at least in my experience as a user, is the fact that Desktop Linux is janky to use. Unless you have the time and patience to learn about all the subsystems and fix the issues that keep cropping up, you're going to want to just spend the money on an OS that "just works". OSX is obviously the best in this regard, but modern Windows with automatic driver installs comes pretty damn close.

(As a developer, the fact that there's no universally agreed-upon standards make it a pain in the arse to support too.)

>Nobody's going to have good UX or fancy onboarding if they can barely find enough resources to keep the project going and fix all of the bugs, as well as handle the technological churn of keeping dependencies up to date etc.

IMO, the main issue with open-source isn't actually finding developers (people love working on OSS) but doing the kind of boring, rigorous QA that you'll see in a commercial firm. Nobody's going to volunteer to reproduce specific edge cases in a printer driver, so it sits broken for decades.

When was the last time you used Linux, 2003?

The only problem with Linux is manufacturers that don't advertise support of their devices and that there are way too many distros so you can't have one ecosystem like you have a windows or macos ecosystem with certified devices.

But if you stick with Lenovo thinkpads, brother printers and bog standard everything running a Fedora linux is like running a macos with well supported devices.

We'd nerd having all kind of hardware sold whose compatibility is well tested and sold under the <insert your favorite distro> brands.

For the 95% use case it's fine but the 5% where stuff just breaks is infuriating.

Office just doesn't work right, and random things subtly break and it's hard to fix even as a technical user. I don't want to navigate between a bunch of files and waste time editing config files for hours. Give me a UI that tells me how to fix my shit.

Only place where I run linux now is on a virtualized server using Proxmox. Linux is a great server OS but it just doesn't work very well as a user distro.

This applies to all OSes.

> I don't want to navigate between a bunch of files and waste time editing config files for hours.

Not something you usually do on Linux but well.

> Give me a UI that tells me how to fix my shit.

My understanding of the UI on proprietary OSes when something needs to be fixed is they don't tell you at all how to fix their shit. You just enter an insane loop of "please wait while we fix your issues" dialog windows which almost never solve the issue.

At least in the FOSS world you can usually look at the logs and understand the error. Open the windows event viewer and you are mostly faced with codes that you have to search online for a signification.

2019, with the LTS version of Kubuntu (18.04 at that time). Moved when Windows 7 EOL was announced.

The hardware I specifically had trouble with was a Brother MFC-9320CW, a Kingston MobileLite SD card reader and a TP-Link TL-WN722N WiFi dongle. All ostensibly supported, all janky to the point where I just gave up and installed Windows 10.

I also had some Intel Atom Laptop with Arch and i3wm. I didn't mind that so much since I only ever used it for a few specific functions and none of them were critical for my usual workflows. It also flew with i3. I do remember both the WiFi and CPU power management semi-regularly fucking up though.

You can be as smug as you want with your "2003" comments, but the reality is that it's simply not fit for purpose as a general purpose consumer OS.

Tell that to my girlfriend who swear against her windows 10 laptop that takes 30 minutes to be usable at boot and who 99% of the time end up borrowing mine to get something done.

Do you call that fit as a general purpose consumer OS?

It’s not just schools don’t want to use open source software and prefer MS Office, the second largest private employee is a huge licensee of Office. I assure you they know a little about Linux…

This is the dichotomy of Open Source. The tool tinkerers can build fantastic things - but only what they want because it is their free time and effort. If it happens to align with the market, then great, if not, too bad.

My personal favourite exception is Blender. As a long running project it seemed to get more and more aligned with the market until the industry saw it was "close enough" to start supporting it.

If we could replicate this for other tools it'd be nice.

No, this is not the point of school.

Maybe we oughta collectively write better open source software if we want to opine on what schools should do? Gimp doesn’t have to suck, but it does. I say that with true love for gimp and admiration for the people who write it, and I can’t complain about it because I’m not helping write it. (But that is an option, you and I could contribute to gimp.)

Part of the problem is that neither of your analogies are relevant in practice. Gimp isn’t a cheaper less automatic Photoshop, it’s a tool that often doesn’t meet professional workflows at all, and certainly wasn’t made kid friendly. It’s not the same tool and can’t do the same things. Also my kids were taught both gimp and photoshop, and they hated gimp. (And adobe makes photoshop available to schools for relatively cheap, compared to retail.)

The analogy about the software being serviceable is not going to fly in schools. Submitting patches is not something teachers can do. Submitting bug reports isn’t either. So what good is a theoretical idea that the software is open? The only thing schools can do in practice is pay money for support. And money is the single biggest problem they have. Vague concerns about privacy and lock-in are just way down on the list of school & teacher priorities, right? I’m wildly in favor of having good open source software, of avoiding education market capture by for profit companies, and having insanely better privacy controls. Open source sounds awesome, but I think it needs to improve before we start demanding that people use it or teach to it.

Wait, let me stop your metaphor right there. How often do you imagine end-user software like Photoshop or MS Office actually "breaks" in a way that actually requires "repair"?

Imagine if you will, that most of the time that your magic screw gun happens to jam, that simply unplugging it and plugging it back in again magically cleared the problem up, and you could continue where you left off.

It's like a world where everyone uses a particular kind of tool that works well and is expensive, but also… the standard. Everyone expects you to know how to use this tool to do a particular kind of common work.

And then you come along and say, no, we should teach people how to use a _different_, cheaper tool! It doesn't work nearly as well, and isn't what people in the "real world" use, but if you spend years familiarizing yourself with a completely different field, you can service and maintain it yourself!

You might not get many takers.

If you don't understand how much more powerful Vim and Emacs are over notepad, most people who do know the difference would see the flaw in your opinion that all 3 are text editors and one is as good as another.

The same is true of Photoshop vs gIMP. Photoshop has non-destructive editing. gIMP does not. That alone is a huge force multiplier. Photoshop has non-destructive layer and group styles, gIMP does not have layer styles as all. That's just few of the 100s of features Photoshop has that gIMP doesn't. They're not just minor features, they're force multipliers and game changers. gIMP has been planning to add the major ones for 15yrs or so but for whatever reasons has not gotten there.

https://www.gimp.org/docs/userfaq.html#when-will-gimp-suppor...

Oh yes. I'm not using Photoshop, but some hidden-secret alternative developed by two brothers from Bavaria (PhotoLine), because I'm only doing some small-scale private hobbyist stuff, and an (even at the time) slightly older version of their software had been included for free in a photography-themed edition of some computer magazine.

That old version didn't have non-destructive layer effects (although even at that time it already supported non-destructive scaling/rotating/shearing of layers, which was quite nice, and which I gather GIMP doesn't support even today?). As my proficiency with it grew, I started running more and more into the limitations of that, until I eventually decided that maybe I should finally just buy a license for the current version of the software, since of the features it had gained in the intervening years was indeed non-destructive editing.

This turned out to be absolutely the right decision, because non-destructive layers are indeed a game changer. No more making lots of backup copies of layers, no more clumsily noting down somewhere in the layer names or wherever what sort of effects I had applied in case I needed to re-tweak something, non-destructive liquifying, etc. etc. Plus a few nifty other features, and still the same familiar UI despite the large version jump, so absolutely no reason to regret the upgrade.

It's only the weaker students who are quick to "demand the latest shiny industry standard gizmos... because, like, jobs". And of course the management twonks who've been blinded by vendor lobbying and kickbacks. Truth is, modern software is so slick any fool can grab a ready-made plugin-o-matic that thinks for you and wipes your arse, but without understanding principles the moment they change the GUI you're completely lost.

This is why teachers use Word and Google Docs. They do not give a shit about the tool. The tool is a means to an end. Word and Docs are familiar. Word and Docs work.

If you care about teaching students to design good buildings, why waste a ton of time showing them how to place screws.

And it’s a different skill set for sure, but highly worth it - someone who knows how do use it can hang drywall by themselves pretty easily.

If that’s your job, you should definitely know how to use it (and how to use just the gun, and the simple drill).

If it's an English Literature course then nobody cares about the tool used. Moleskine, Google Docs, Word, whatever, who cares. Google Docs is free. Why give yourself more hassle than that?