Hi, I'm Shiloh, the guy that made the original reddit post and medium article. After reading some comments, I clarified my statement on GitHub's role in the CS industry
I noticed the more prominent "Purchase" color on the upsell plan. That either means you went above and beyond on your design research, or have better instincts than many founders.
Good work! Keep it up and you'll really go places.
It would be great if you could re-draft that clause in a way which makes sense, and then get back to your teacher to see if she can influence the competition organisers. That way, next year's competition will be just a little fairer.
The reality is that tools like send-email just... don't really work under the current email oligopoly (since the big providers both heavily filter emails and generally consider IMAP to be insecure). That entire model dates back to the days where email accounts mostly were tied to universities and as a result an emails name had a lot more meaning.
Heck, even for the kernel, send-mail is starting to fall out of practice because it just renders like shit in most email clients. It's a rather dated model.
The pull request flow is the de-facto industry standard, basically all but the oldest git-based projects use it. Not all are on GitHub (since it's proprietary or just straight up issues with GitHub as a company), but even when they aren't, self-hosted git servers like gitea and gitlab are both still using that exact flow.
The examples where this isn't the case are generally rare enough that they're worth more learning as a procedure for that project than as an easily transferrable skill.
It's not realistic to suggest people avoid GitHub with the kind of market dominance that GitHub has. I while I empathise with the sentiment learning GitHub is absolutely a tool that will help them get jobs in a way that Git alone won't.
How many advanced GitHub features not present on other platforms do you think students are using? If the answer is none, then there's no reason to choose a proprietary service they can't hack on and are required create accounts with a for-profit entity.
We shouldn't use Keybase, but the nearest alternative open-source, Keyoxide, is not functional unless you're an engineer. Keybase on the other hand is perfectly usable without being an expert in the underlying cryptography.
Similarly Github has a huge number of features that a usable by people that aren't developers (PRs etc).
If you want a point to your side, GIMP is a bad hill to die on. Blender is a much better example of a tool that we should be teaching as opposed to Autodesk Maya (etc).
That's the thing with Blender though, it has feature parity with the things it's competing with and a community and network effect. For something open source to compete with GitHub it would need to be so much more than Git and offer the community and discovery that forms a big part of the "open" GitHub.
In the meantime people need to get things done. NOT teaching them the tool they're inevitably going to use is just a pointless battle to fight. Better to create the alternative first and lure people with straight up feature benefits.
Blog has been under redesign-rebuild for a long time--other priorities. Said services were planning on being removed and replaced with alternatives as mentioned.
Blender is a better option I agree.
We don't disagree in principle about open tools being better. I do however think that there isn't a viable alternative to GitHub at this point. The same could have been said about ExpertExchange at one point, and while StackOverflow isn't perfect, they do at make everything Creative Commons.
Tackling GitHub would require building a platform where all of the other tools such as Pull Requests, Actions, Issues were also represented in a portable format. Recognising that Git is just one interface to GitHub now rather than the main part of it.
Yes, there is no 100% feature parity. But there is feature parity in the things that people actually use.
The Adobe Creative Cloud ecosystem has very strong lock in todo with workflow and collaboration (amoung other things). It's not that you couldn't do the vast majority of those things with another tool (open or not, e.g. Affinity), it's that the company you're going to work for is going to look at your Resume for experience with Adobe, and issue you a laptop with adobe on it, expect you to produce AI and PSD files (and now of course Figma for certain subsets).
GIMP is clunky and is not pushing the boundaries. Some of the newer smart tools in Photoshop are impressive in their ability to select and delete objects. Even Affinity which offers very polished products is going to struggle.
Blender on the other hand is a prime example of where an open tool can make strides to break the stranglehold. It has a very focused team of collaborators and a clear roadmap that is taking it from a tool that was quite hard to learn even if it was powerful into something that could become the industry standard.
The UX of Blender compared to what it was 5-10 years ago is so much better. I'm not sure the same could be said of GIMP.
That power comes both from strong direction and a growing and active community.
It's not close to being wholly about the capability of the product itself.
Excel has the incumbency advantage - more people are familiar with it and more people already have it. Switching to LibreOffice therefore often means switching to a tool you don't know as well and then living with interop issues with your co-workers that are still on Excel. For LibreOffice (or any other competitor) to do something with "similar effort", it has to overcome those headwinds.
If you look at the history of spreadsheets, there is a long history of "better" products coming along, but only a few that were good enough to be able to displace the incumbent.
* VisiCalc - The First
* Lotus 1-2-3 - Runs on the PC, has graphics, and a bunch more capacity than VisiCalc
* Excel - GUI support
* Google Sheets (maybe) - Online real-time collaboration, etc.
There were many others along the way, and many of them could make the same claim you're making here about LibreOffice. (Multiplan, Improv, WingZ, Quattro Pro, Framework, etc.....)
Parity is ok, but the goal really has to be 'so much better that it justifies all of the switching costs'....
Will disagree on that for as long as GIMP lacks non-destructive editing (according to FAQ will be introduced in v3.2) since that changes drastically workflow.
It does. GitHub won. Git and GitHub are synonymous now.
Arguing decentralized version control, like Git, should have a centralized home is antithetical to the tool.