Windows Under the Hood

Here's the reassuring truth: Windows is doing the exact same job every operating system does. A kernel in the middle, sharing the hardware among a crowd of programs, deciding who's allowed to do what. If that sentence is fuzzy, the What an Operating System Is guide builds it from scratch - this phase assumes it and shows you what it looks like wearing Windows clothes.

What trips people up isn't the model. It's the local dialect: drive letters, a folder layout that scatters your stuff across several places, a permission prompt that darkens the whole screen, and a few decisions Microsoft made decades ago that still shape everything. Let's translate.

The NT kernel: the engine you never see

What it actually is. At the center of Windows is a kernel called NT (the name is a holdover from "New Technology," Windows NT, 1993). Every version of Windows you've touched since Windows XP - including Windows 10 and 11 - runs on a descendant of that same NT kernel. It's the trusted core that talks to the CPU, the memory, and the disk, and enforces every rule about who can do what.

📝 Terminology. NT kernel = the core of Windows that manages the hardware and has full control of the machine. When people say "the Windows kernel," this is it.

Why this matters to you. Everything you see - the desktop, File Explorer, the Settings app, your browser - runs on top of NT, in user space, asking it for anything that touches the disk or network. A crashed program rarely takes the whole machine down, because NT just cleans it up and moves on; a true hard lockup means the kernel itself is in trouble.

💡 Key point. "Windows" is two things wearing one name: the NT kernel (the real operating system) and the desktop shell (the Start menu, taskbar, File Explorer - the face you click on). When the taskbar freezes but music keeps playing, that's the face hanging while the kernel underneath is fine.

Drive letters: C:\ and friends

What it actually is. A drive letter is Windows' name for a volume - a formatted area of storage the OS can read and write. C: is, by long convention, the drive Windows itself is installed on. Plug in a USB stick and it shows up as D: or E:; a second hard drive might be D:. The letters aren't magic - they're just labels the OS hands out.

Why people get this wrong. People assume drive letters are physical - "C is my hard drive." Not quite. One physical disk can be split into several volumes (C: and D: on the same drive), and several disks can each get their own letter. The letter names a volume, not a piece of hardware.

📝 Terminology. Volume = a formatted, usable storage area with a filesystem on it. Filesystem = the scheme that organizes files and folders on a volume. Windows almost always uses NTFS (NT File System) for its main drive.

How this differs from Unix. This is the first big fork from the macOS/Linux world. On Unix-like systems there are no drive letters at all - there's one single tree starting at / (the "root"), and extra disks get mounted into a folder inside that one tree. Windows keeps each volume as its own lettered tree:

   WINDOWS                          UNIX (macOS / Linux)
   ─────────────────────           ─────────────────────
   C:\                             /
   ├── Windows\                    ├── usr/
   ├── Program Files\              ├── etc/
   └── Users\                      └── home/
   D:\   ← a separate tree         /mnt/usb   ← the USB lives *inside* the one tree
   E:\   ← another separate tree

Neither is wrong - they're two answers to "how do we name more than one disk." Windows says "give each its own letter." Unix says "everything hangs off one root."

The folder layout (and AppData, the one everyone trips over)

Open C:\ and you'll see a handful of folders that actually matter. Here's what each is for:

   C:\
   ├── Windows\          ← the OS itself: the kernel, drivers, system tools. Leave it alone.
   ├── Program Files\    ← installed apps (64-bit). One folder per app.
   ├── Program Files (x86)\  ← installed apps (older 32-bit). Yes, two folders. See below.
   ├── ProgramData\      ← app data shared by ALL users (licenses, shared databases). Hidden by default.
   └── Users\
       └── you\          ← your home folder. Everything that's "yours" lives under here.
           ├── Desktop\, Documents\, Downloads\, Pictures\ …
           └── AppData\  ← your per-user app settings and data. Hidden by default. ← the famous one.

Why there are two Program Files folders. Program Files holds modern 64-bit apps; Program Files (x86) holds older 32-bit ones. Windows keeps them apart so a 32-bit and 64-bit version of the same thing don't collide. It's not a bug - it's Windows quietly running both worlds at once. The x86 name refers to the old 32-bit Intel architecture.

Now, AppData - the folder that explains a hundred mysteries. It's a hidden folder inside your home folder, and it's where applications stash your personal settings, caches, profiles, and save data. When you ask "where did my app save its config?" or "where are my game saves?" or "this program forgot all my settings after I reinstalled" - the answer is almost always somewhere in AppData. It has three sub-folders, and the difference is worth knowing:

   C:\Users\you\AppData\
   ├── Local\      ← machine-specific data. Big caches, this-PC-only settings. Does NOT roam.
   ├── LocalLow\   ← like Local, but for low-permission apps (e.g. stuff running in a browser sandbox).
   └── Roaming\    ← settings meant to follow you between PCs on a corporate network. Smaller, portable.

The split exists for a real reason: in a company, your Roaming profile can follow you when you log into a different computer, while Local (full of huge caches) stays put. On a home PC you'll mostly care about Local and Roaming being the two places apps hide things.

A real example. You can jump straight there. In File Explorer's address bar, or anywhere a path is accepted, type %AppData%:

C:\Users\you> echo %AppData%
C:\Users\you\AppData\Roaming

What just happened: %AppData% is an environment variable - a named shortcut Windows expands into a real path. It points at your Roaming folder. (There's a matching %LocalAppData% for the Local one.) This is why guides say "go to %AppData%\SomeApp" instead of spelling out C:\Users\yourname\AppData\Roaming\SomeApp - the variable works no matter what your username is.

⚠️ Gotcha: AppData is hidden by default. If you go looking in File Explorer and can't find AppData, it's not missing - it's a hidden folder. Turn on View → Show → Hidden items, or just paste %AppData% into the address bar and skip the hunt. This is the folder people mean when they swear an app "deleted all their settings" after an update - it didn't, they just never knew it existed.

Processes and permissions: what "Run as administrator" really does

You met user space and kernel space in the OS guide: untrusted programs run in user space and must ask the trusted kernel for anything powerful. Windows adds one more layer within user space, and it's the thing behind every "Do you want to allow this app to make changes?" box.

What it actually is. Even when you're logged in as an administrator, your normal programs run with limited privileges - as if you were a standard user. This is on purpose. It means a random program (or some malware you clicked) can't quietly change system files just because you happen to be an admin. When a program genuinely needs elevated power - installing software into Program Files, editing Windows\, changing system settings - it has to ask, and you have to approve. That approval prompt is UAC.

📝 Terminology. UAC (User Account Control) = the Windows feature that keeps programs running with limited privileges and pops up a consent prompt when one needs to do something powerful. Elevation = the act of granting a program those higher privileges for the task.

   Standard rights (default)            Elevated rights ("as administrator")
   ─────────────────────────            ────────────────────────────────────
   • read most files                    • write into Program Files\ and Windows\
   • write inside your own Users\ folder • install / uninstall software
   • run normal apps                    • change system-wide settings
   • CANNOT touch system folders         • edit other users' data
        │                                       ▲
        └────────── UAC prompt: "Allow changes?" ┘
                    you click Yes → the program is elevated for that task

What it does in real life. Right-click a program and choose Run as administrator, and Windows checks with you via the UAC prompt (the screen darkens to a "secure desktop" so a sneaky program can't fake your click). Approve it, and that program now runs on the powerful side of the line.

Why this saves you later. Half of "it won't let me save here" and "the installer failed" problems are this. Saving into C:\Program Files\... and getting "Access denied"? That folder needs elevation, and your editor isn't elevated. A terminal command fails with a permissions error but works after you reopen the terminal as administrator? Same lesson - you're on the wrong side of the privilege boundary, not fighting a random glitch.

⚠️ Gotcha: don't run everything as administrator "to be safe." It's the opposite of safe. UAC's whole value is that your everyday programs can't wreck the system even if they misbehave. Run a thing elevated and you hand it the keys. Elevate only when a task genuinely needs it.

How Windows really differs from Unix

You'll hear "Windows is different from Mac/Linux." Here's the straight, concrete version of how - the differences that actually bite you, not vibes:

Two of these are worth a sentence more, because they're the ones that surprise people:

• .exe and extensions. The file extension isn't just a label on Windows - it's how the OS decides what a file is and how to open it. Double-click report.txt and you get Notepad; rename it report.exe and Windows will try to run it. That's why hiding extensions is a security risk (invoice.pdf.exe looks like a PDF when extensions are hidden) and why Windows asks "are you sure?" when you change one.

• The registry instead of config files. Deepest cultural difference of the bunch, and it gets its own section in Phase 2. Short version: where a Unix program drops a text file to remember its settings, Windows often writes them into one giant, central database called the registry - powerful and fast, and why "just edit the config file" isn't always how you fix things here.

💡 Key point. None of these differences make Windows "weird" - they're just different design choices for the same problems. Drive letters vs. one root. Extension vs. permission bit. Central database vs. scattered text files. Once you can name the choice, the difference stops being friction and starts being something you can reason about.

Recap

1. Windows runs on the NT kernel - the trusted core. The desktop, File Explorer, and your apps all run on top of it in user space.
2. Drive letters (C:, D:) each name a volume, not a physical disk; Unix instead hangs everything off one root /.
3. Apps live in Program Files (and Program Files (x86) for old 32-bit ones); your settings and data hide in AppData - a hidden folder with Local, LocalLow, and Roaming inside.
4. UAC / "Run as administrator" is your programs asking to cross from limited to elevated privileges - the source of most "access denied" mysteries.
5. Windows differs from Unix in real, nameable ways: \ vs /, drive letters vs one root, .exe/extensions vs the permission bit, and the registry vs config files.

Next, we'll watch the parts that run without you ever opening them - services - and meet the registry up close.

← Guide overview · Phase 2: Services, Task Manager & the Registry →