OSDev.org

On this page, https://wiki.osdev.org/MZ, at the PE extension section, the following paragraph:

(Actually, that's incorrect – this extension was originally added in the New Executable format, which was first released as part of Windows 1.0 in 1985, and the multitasking MS-DOS 4.0 in 1986. The "PE header start" is called "e_lfanew" in Microsoft's headers, since it was originally the pointer to the NE header, although later was also used for LE, LX and PE; in principle it can be used for *any* executable format which is defined as an extension of MZ.)

Actually, that's incorrect
*any*

I would like someone to check this part.

Why would those phrases be suspicious? I don't understand. But in the future, if someone finds something that's incorrect, one should just correct it, rather than commenting that it's incorrect.

The two parts you quoted don't suggest AI/LLM to me. However, I do find the '–' character suspicious (which differs from '-' and looks a typical LLM-ism, though that would usually be '—' -- so many dashes!).

I went through the author's edit history (History -> https://wiki.osdev.org/index.php?title= ... ldid=28644 -> contribs) and his edits don't seem out of the ordinary. I can't verify if his edits are accurate though, as I'm unfamiliar with the MZ format and floppy disks.

I am also not very familiar with the MZ format, that is why I was reading the wiki page.

It is a rare pattern for LLMs, but sometimes they write something and then correct themselves in the same reply, especially in complex/old topics "Actually that's incorrect -" pattern. Also, the "*any*" part is clearly Markdown format for writing bold font, clearly not the same format used in this wiki, again when copy-pasting from an LLM's response this can happen.

Even if it's not AI generated, why would the wiki page incorrect information and then auto correcting itself afterwards?

josufh wrote: ↑Thu Dec 04, 2025 8:09 pm I am also not very familiar with the MZ format, that is why I was reading the wiki page.

It is a rare pattern for LLMs, but sometimes they write something and then correct themselves in the same reply, especially in complex/old topics "Actually that's incorrect -" pattern. Also, the "*any*" part is clearly Markdown format for writing bold font, clearly not the same format used in this wiki, again when copy-pasting from an LLM's response this can happen.

Even if it's not AI generated, why would the wiki page incorrect information and then auto correcting itself afterwards?

Because the Wiki has more than one author. Look at the history. Sjk, who made the last 3 contributions, in Feb 2024, added the (actually that's incorrect...) comment. The wrong comment about MZ being introduced with PE? That was written by Unprotectedmode on 8th September 2020.

I spotted an obviously LLM-generated edit, but it doesn't look malicious so I've approved it anyway.

https://wiki.osdev.org/index.php?title= ... oldid=5921

I'm tempted to revert it though since it reads more like an advertisement than providing actually useful information like data layout.

I am author of the xfs and befs edit.

Sorry, I am native japanese speaker and use AI to help format/translate my work.

Going forward I will move "increment" forward with clean verify info and make small edit so my hard work not all burn by one or two bad comment or AI translate issue.

My English... not good enough for write direct technical personal computer article (T.T)

Salimanz23 wrote: ↑Tue Dec 30, 2025 11:08 pm Sorry, I am native japanese speaker and use AI to help format/translate my work.

That explains it The edits clearly have positive intent but they read quite strange.

The BeFS edit in particular bothered me because of a few words/sentences (emphasis mine):

Unlike many contemporaneous filesystems, BeFS treats files not merely as byte streams, but as structured objects with indexed attributes, enabling database-like queries directly at the filesystem level.

Attributes are not merely passive metadata.

"merely"? The use of the word isn't wrong, but is unusual to see it used here.

From an OS design perspective, BeFS blurs the boundary between storage and information retrieval.

Nevertheless, BeFS remains influential. Its ideas—particularly filesystem-level indexing and rich metadata—anticipated later developments in desktop search and semantic filesystems. Open-source reimplementations exist (such as those used by [[Haiku]], BeOS’s spiritual successor), preserving BeFS as a notable example of filesystem design driven by user-level responsiveness rather than purely by storage efficiency.

This sounds like it comes straight out of a marketing pamphlet

Demindiro wrote: ↑Wed Dec 31, 2025 8:35 am
That explains it The edits clearly have positive intent but they read quite strange.

The BeFS edit in particular bothered me because of a few words/sentences (emphasis mine):

My apologies for the trouble. I will fix these articles - and use small edits so my work isn't destroyed by mistakes.

(this part AI Translated because my English)

I fully understand the problem as explained in your post and messages on my discussion page. I used AI inappropriate to the purpose of the wiki.

While I cannot write articles directly in English, I will make sure my edits are specific, useful and in small pieces so parts can be reverted without breaking all of my hard work.

I only intended to fix these articles because of just how stubby they were and basically offered no useful info,

(end AI)

Sorry for the trouble again!

edit: All edits for deep technical explanation will be cited specifically. I rather "over cite" page, than have my work deleted.

Final Statement on XFS

XFS article is uh, "crystallized" I think? I share on-disk structure, in future when I have time I will specifically go into differences in formats and structures of the different "version"

Now to fix BeFS.