Monday, July 20, 2026

Human questions to ask in code reviews

Automated (AI) powered code reviews are becoming increasingly popular and powerful. I use them myself to catch things I miss.

But there are questions I ask as a human reviewer that I've yet to see any AI-powered tooling help with.


 Questions like:

  • Does this fix the cause or the symptoms?
  • Are there related changes that should be addressed while we're in this part of the codebase?
  • Is this just fixing a problem or actively making things better for the user? (They aren't always the same.)
  • Do we have related docs or samples that need correcting/updating or adding to?

Wednesday, July 08, 2026

Is second-order enshittification a thing?

The Enshittification poop emoji, created by No Ideas' Devin Washburn for the Farrar, Straus and Giroux edition of "Enshittification: Why Everything Suddenly Got Worse and What to Do About It" (Cory Doctorow, 2025).

 
Enshittifcation, as a term popularized by Cory Doctrow, has for many become a catch-all term for things generally getting worse. I've found it useful (unsurprising if you know me) to think about it in two categories.
There's the popular definition of how very large, dominant platforms change [to get worse] over time as business/organizational priorities change.

I think there's a second-order version of this that follows as a response.

I'd sum it up like this:

"If [big company] doesn't have to focus on quality, we can cut corners there too."

What happens when "everyone" takes this approach is that things get continually worse. Not all at once and not always in the immediately obvious ways. More in the million paper cuts way.

From a software perspective, it's the increasing number of small bugs that no one seems to care about fixing. This teaches people to not bother reporting them. After all, "what's the point? They're not going to fix it even if I [you] do report it."

With fewer known or reported bugs, the company thinks everything is fine, or worse still, thinks skipping on quality is an approach that works.

And a self-reinforcing cycle continues.


I prefer to think of it another way.

Software exists to help people and to make their lives "better". This could be in one or in many ways.

When people are constantly being interrupted, frustrated, and disappointed by the software they use, it's very hard to argue that it is making them better, happier, or more productive.

This is part of the reason I think that QA, customer service, and fixing bugs are among the most important things that those in software development can focus on.

With AI making copying software easier, it's focusing on how it relates to people that can be the real differentiator. 

Yes, AI may mean you can vibe code a copy of something that exists, but does it include support for all the non-obvious edge cases? Do you even know all the features that exist in what's being copied? How (and?) will you handle any bug reports?

Yes, building software can be fun for you, but high-quality software that works reliably benefits many more people and, in a small way at least, makes the world a better place.


And this can also lead to mediocrity

A small company tries to compete with a large one. They try to do everything because that's what the large company (or their software) does. They try hard, but as they're stretching themselves so thin, everything suffers. "Never mind", they say, "at least we're trying our best." But, for what? Who benefits? The company's staff exhaust themselves producing a mediocre product, and it's no better than the alternatives.
Trying to compete with a large company and basing goals on the perceived [low] standards of another company doesn't produce a great product, happy customers, or happy staff.



_This was written many months before posting--to avoid anyone thinking I'm responding to a single person, company or incident_


Wednesday, November 12, 2025

Resetting the experimental instance of Visual Studio has changed in 2026 (v18)

Updated Jan 7th 2026 -- see below


If you build extensions for Visual Studio, you'll know that sometimes (frequently?--depending on what you're doing) it's necessary to completely reset the experimental instance to get it back to a known/good state.

With previous versions, it was possible to start typing in the main Windows menu and it would find the command to do this for you.


This was possible because the menu included entries for just this command:

(Note that I have multiple entries because I have multiple long-term support versions installed for testing.)

However, VS2026 does not include these menu entries. (Who knows why they were removed...?)

Instead, you can (must) trigger the resetting of the experimental instance from within Visual Studio from within the Feature Search. Simply start typing the command in the search box, and it'll find it for you.


Note. I couldn't find another way to access this functionality in VS2026. If you know of one, please share it.


UPDATE - Jan 7th 2026

I can no longer get the above to work, as the feature search no longer finds the option.
I asked Copilot and it said the only way to do it was via the command line.


Take this with a pinch of salt though, as the above includes incorrect information about the version number as VS2026 is version 18!



Wednesday, August 27, 2025

Is writing a test a good contribution to an open source project?

The cliché always used to be that "contributing to docs was a great way to get started in open source." Now I'm also starting to hear people suggest that writing a test can also be a good entry to a project.

But is this a good idea? I'm not sure...

"writing hand" and a test tube
Say you have a piece of code that isn't covered by any tests. It's fair to say this isn't an ideal situation to be in. All things being equal, having tests for this code would be a good (better) thing.

But not all code and not all tests are created equal.

Is adding a test for a piece of code that is never expected to change in the lifetime of the project valuable?

Is it valuable to write tests for code that is so clearly understandable that if anyone changed it, then lots of things would obviously be wrong, and a manual review of the code would easily spot the problem?

Is it valuable to add tests for only some scenarios or paths through a piece of code? Sometimes. Sometimes not.

Is adding tests that ensure all possible input can be handled by the code a good addition? Maybe, but if the project has been around a while, then all such inputs have likely been encountered already. If there were inputs that might cause a problem, they've most likely been encountered and dealt with.

You may be able to create a lot of tests very quickly. (Especially if using AI.)  But is it worth running them? If they don't run quickly, is it worth the delays and the money/energy it takes?

Coded tests must also be reviewed like any other code contribution, and reviewing PRs is a common bottleneck in many OS projects. 




I'm not against tests.
I think automated tests are great, and everyone should write more of them.
I just think that adding them after the fact is the wrong time to do it. It's harder to do it well, and they risk being low value.
Writing (or at least documenting) all the required tests before you start coding is the best time to write them.

Of course, if there's a project with documented manual test steps and you want to write code to automate them, then that sounds like a very valuable contribution. (Just as long as it doesn't require modifying the underlying code to make that possible.)

Or, if you want to help with the testing of a project, look at some open issues and start documenting how to test those features when they are implemented.

As with any open source project, the best kind of contributions are the ones the owners and maintainers are asking for, and if they're of any size, they never start with a PR but with a discussion or issue.

Thursday, August 07, 2025

Miscellaneous AI-related questions

Question mark with AI-sparkle

No answers. "Just", questions I'm aware of and considering:

  • If working with AI means communicating with machines more like we do with other humans, how do we avoid things also going back the other way and treating people more like machines?
  • Are "agents the future of [all] work"? And, if not all, how to identify the work that can change or be replaced?
  • If "AI is only as good as your data", why isn't there as much effort being put into ensuring the quality and accuracy of the data as there is hype about AI?
  • At what point does AI not need human oversight? All the education highlights human oversight, but the futurists don't include it...
  • What is in the middle-ground between traditional GUIs and "just" a text box?
  • As feedback is highlighted as being essential when developing tools with AI, is there a way for feedback from a tool to be passed back to those creating the underlying models?
  • If there's a GUI for something, does it automatically need (and benefit?) from having an equivalent interface that's accessible via command line, API, and Agent/MCP?
  • As speed/rate of change is a common complaint among all types of people and people doing disparate tasks, how do you factor this in when introducing AI-powered tools?
  • If people are generally reluctant to read instructions, why will they happily read the text-based response from an AI tool telling them how to do something?
  • Asking good questions is hard. How people ask questions of AI-powered tools greatly impacts the quality of results. In training people to use AI, are they also being taught to ask good questions?