Hiding file extensions is uniquely dangerous on Windows as the file extension is the canonical reference of what type something is. Change the .doc to .exe and the system will try to treat it as an executable. At least on other systems it's a convenience and the system identifies type and executable bit in somewhat more robust ways.

Yes, that's true too. It's a common Windows malware distribution vector - bad actors send files with names like "photo.jpg.exe" or even "photo.exe", with the icon resource in executable's metadata set to look like default image icon, or even a thumbnail of some real photo.

Not only on Windows, on other OSes as well.

Many developers don't rely on the OS API to check the file contents and use a simple extension checking.

Countless words have been written about this, but does MS see it as a problem?

Yep. It's about making things feel simple and easy, without actually being simple and easy.

My parents both started using word processors in the days of MS-DOS and they had no trouble grasping the idea of files and directories, or that a WordPerfect file ends in `.wpd`.

>>Hierarchical folders and associating files to programs via name extensions isn't that hard to grasp.

Actually a lot of people seem to struggle with it, I used to run training courses on a product that had a tree based configuration. As soon as you got beyond a basic list and started building a hierarchy half the class would fade away.

> half the class would fade away.

They didn't care, but every one understands it. Then there is also the basic belief that everything to do with a computer is hard, so they've already decided they can't understand it.

Tax returns > home > By year > (this number is federal, this other number is state)

Tax returns > business > By year > (this number is federal, this other number is state)

We use standard filing and organization like the a basic file system all the time, people have decided that it's hard on a computer.

Compare it to some trees they know from real life. Company organizational chart. School organizational chart (principal has teachers who have children). Country organizational chart (president over ministers over ...).

Trees are not a trivial abstraction, but they aren't also hard.

Well I don't do it any more but at the time I tried lots of different approaches, some people just don't seem to have the mental capability to handle the abstractions.

That's amusing actually. Average HN reader just cannot imagine an average people cannot handle abstractions at all.

If an average human couldn't handle abstractions at all, their brain would explode somewhere around being 2 years old. Everyone, even the biggest luddite, deals with shit ton of abstractions every day. The brain can handle it. The ego, quite often, cannot - and herein lies the role of a good teacher.

> by catering to the use cases of the common man, you're limiting them to your interpretation of what the common man's use cases are.

Or, using telemetry, user studies and more, you can actually figure out what those use cases are. It's not rocket surgery.

> Or, using telemetry, user studies and more, you can actually figure out what those use cases are. It's not rocket surgery.

Or you end up convincing yourself into thinking your own preconceptions are actually the real use cases. It's like social science, except most software development companies haven't heard of any kind of scientific rigour. Not to mention the study tools are broken on purpose too; it's not unheard of that people writing e.g. analytics / A/B-testing tools optimize them for the "wow it works" aspect - successfully, because most companies using them don't know squat about stats.

That said, even if you identify the right use cases, there's the orthogonal aspect of whether your tool will allow for those and only those use cases, or if it is a more general one.