Computing should enable humans: how to choose tools

March 02, 2023 12 minute read

the first thing our new hire did was fix a bug that's been bugging him forever as a user prior to joining.

he then breathed a sigh of relief and submitted his two weeks' notice. wtf??

Tweet by @swaglord__420 on Mar 30, 2021

In celebration of the fixing of the notorious 4 year old VSCode bug which practically everyone I know has seethed about where diagnostics show below pages of function documentation, I think it's important to write about choosing tools that support and enable humans by aligning with our values.

In order to enable humans, tools must be reparable. If you find a fault in the tool, it must be possible to get it fixed in a timely fashion. Imagine having a broken stove that only works if you jump twice, spin around three times, then smack the thing, only to have a 50% chance of turning on the burner, but your landlord won't do anything about it. Every day you use it, you expire an exasperated sigh, hoping it will work this time.

The same is true for software: either you need to be able to replace it, fix it, or get it fixed, but all need to be able to happen in a timely manner. Unfortunately, everyone will find bugs, and sometimes, they are sufficiently problematic that they get under your skin every single day.

As someone with a troubled upbringing, I am intimately familiar with the feeling of lack of agency over my life, and I'm determined to never have that happen again in any facet of my life. For the most part, I feel that the computer generally gives me more agency, but that is not the case for everyone, not every developer, and absolutely not every non-developer.

The way that a lot of people experience software is that it comes from magical beings from on high, and they pray that it works correctly, for they can't do anything if not. We need to have empathy for people who use our tools: as the people from on high bequeathing the rest of the world with tools, we are in a position of power over them and their lives.

We have constructed a world where software serves to take agency away from people: if people are forced to use your tools, you have a responsibility to them to make it usable. That means accessibility; that means that you are accountable to your customers when bugs are reported; that means that your system's decisions can be appealed when they have influence over lives.

Structural factors

Various structures of software development have different effects on whether the software gets fixed.

Open source is not necessarily the answer. We see that that VSCode bug has been present for four years, that the fix was 5 lines, that it was submitted by someone outside Microsoft, and that the fix was not reviewed for multiple months. Responsible closed source developers can be less frustrating to deal with than hegemonic open source landlords such as Microsoft who gather goodwill by producing a good open source product, then don't fix their bugs or review their patches, then turn around and close the source of language servers.

A major differentiator of whether you are hosed if you run into a bug or not is whether the bug tracker is public. That is: if there is a problem, then can you see it? I ran into a frustrating bug in a JetBrains product's code formatter, and while it was not being fixed, the bug was at least public: I could see that I am not the only one having it.

Microsoft has a closed bug tracker, although there is the "Feedback Hub", which my experience has been that it is where I find that my bug is a year old and unfixed. Apple has a closed bug tracker, and damningly, community members have made a web site that publishes their own bugs so that they could determine if they were alone.

Another factor that determines whether you are in for misery when software breaks is the release schedule of the software. Assuming that someone (whether you or not) does fix the issue, how long are you going to wait to get the patch? Is this months? Weeks?

Having a large amount of functionality maintained by one group can be troublesome. Is it one giant repository where everything goes into a black hole unless you work there? Which organization owns the majority of the functionality you rely on? Are teams autonomous? How much power does a single entity hold?

Thinking back on why I switched back to Neovim exclusively as an editor, one major reason is the amount of the functionality that is in plugins instead of in the core. Autocompletion is in a plugin. Highlighting queries are in a plugin and can be overridden in your config file. Much of the UI is in a plugin. I can hack up basically any plugin in my config by doing various things to override it without having to fork it. None of these plugins are owned by the same team, and there are alternatives for each piece, so that if it sucks, I can hit da bricks.

Skeleton meme: "just walk out, you can leave!!!" If it sucks ... HIT DA BRICKS

Business model

Consider the business model of any software and services that you use. Is it a two-sided market in which the company has inserted themselves in the middle to extract value? How do they make money? If they don't obviously make money, how might they make money in the future, and is that method acceptable?

The products I am least concerned about using are the ones which have an obvious business model: they have customers, who pay them for something they do.

Consider the case study of Microsoft Visual Studio Code, which is free. How are they extracting money? God knows, but they sure are eating market share and gaining themselves a tremendous position of power in the market by doing so, which could be easily leveraged to do anticompetitive behaviour, advantaging their other products. From the very beginning, it was released as a closed source distribution of an open source tool, and various very important pieces depend on closed source code, so that Microsoft retains full control. For example, the remote extension, an increasing number of language servers, the extension store, all are closed source components under EULA.

As another example, Microsoft Bing tells you to not install Chrome when you use Bing to google "Chrome". Then, Microsoft Edge injects pop-ups into Google's webpage for downloading Chrome. Then they do it again when you try to change default browser in the settings app. This is so obviously anticompetitive that it should have every country suing them for billions for the exact same reason they got sued for billions for doing less to IE, but apparently we do not live in a world where laws apply to corporations anymore.

There are infinite stories of Microsoft abusing monopoly positions in every market they are in, but I will digress.

Venture capital funded startups are crooked. Exercise serious caution in using their products, since they have an obligation to their investors to screw you over eventually.

It's a traditional practice in VC-funded startups to enshittify products, a form of bait and switch: gather a huge market share by providing a decent service at a loss, driving competitors out of business, then, once the market has been burned to the ground, start screwing everyone over for even more money because they can't leave. First, make suppliers depend on the startup by eliminating their agency: Facebook, for example, prioritized posts that did not direct people off-platform, demanding full text of articles, destroying their website audiences. Then, they started demanding money from said news outlets. The news outlets can't say no because their control over their destiny has been systematically destroyed.

Uber destroyed the taxi business everywhere by providing a service for less money, more conveniently, at a massive loss, by subsidizing their two-sided market. They drove the local businesses out of business (and, to be clear, small businesses are often terribly abusive environments, but I would rather them over Silicon Valley landleeches). Then, they started screwing their customers and their employees (sorry sorry, "independent contractors"), extracting maximum profit by inflicting maximum misery.

It's not trivial to blame startups for doing this, because those were the conditions they agreed to when they accepted the money in the first place: hockey-stick growth is the expectation, and their leaders will get fired if they don't do it.

"Software" (venture capital) is eating the world.

Can you patch it?

In the case where you are in the software developer class, and you have free time, it is important that you can patch the software as easily as possible if it has bugs or oversights. This is significantly enabled by systems that reduce the power that others hold over the software: patching, plugins, and extension points. Some of the stories here are based on real life, and the names of the projects in question have been omitted to protect the guilty.

There's various considerations of community health that matter to the selection of software that has influence over your life, such as Linux distributions, programming languages, etc, where you expect to spend a lot of time in that community. For example:

The ease of which you can set up a development environment for the software, fix the bug, and submit the change matters. I share the opinion that mailing lists and git send-email are obsolete gatekeeping nonsense that should be avoided when possible, but it is not the only thing that matters: how fast do patches get reviewed? How functional is the development workflow upstream?

How forkable is the software? If you have to hold a patch against it, how bad is that going to be? A big reason that VSCode is so troublesome to use as someone who wants working software is that it is aggressively not forkable: you have to get stuff merged for it to get into the closed source distribution, they are bad about merging things, and it moves pretty fast while being this massive blob of things that aren't extensible.

In spite of being a card-carrying anti-FSF person, I very much care about so-called "software freedom": I care that I can fix my own software. For example, this is enabled by Nix making patched versions of software no harder to deploy than upstream versions, which unties me from their release schedules when required. Further, there is approximately no difference between package definitions I write and ones that are in nixpkgs upstream, which is a huge boon which combines well with nixpkgs not being insular: I can contribute to it just as much as anyone else. The end result of this is that my operating system no longer holds a significant position of power over me being able to get things done.

Another thing that's useful is when the software is extensible, since it gives you agency as the user to be able to rectify oversights without even maintaining patches. Nix is not without criticism on some of the bases above: I wrote the Nix plugin included with nix-doc to rectify the oversight of upstream not completing their patch for a way of getting function documentation interactively from nix repl. Because it has plugins, I can experiment with new features that aren't yet upstream, and I have accordingly enjoyed having this feature for a year and a half, without worrying about when it will get implemented upstream.

Conclusion

We spend a huge amount of our lives using software, and there are a lot of factors that can make the difference between being under the thumb of software and software being a force of good in our lives. I have been thinking about a lot of these things subconsciously for years and picking projects based on the right vibes, and it's become clear that these are actually values that I care about, not just software things.

Resiliency is important, and it is especially so in things that we need to use to do our jobs and live our lives.

I hope that you too can get to a place where the Computer generally works and is a positive influence on your life.

🏴