Published 15 August 2026 · Last updated 15 August 2026
Keyboard ghosting is when your keyboard reports a keystroke you did not make. Press Q, W and A together and a phantom S appears in your text. That is the literal meaning of the word, and on a keyboard built in the last fifteen years or so it is genuinely rare.
What people almost always mean when they say "my keyboard is ghosting" is the reverse: they held three or four keys and one of them simply never arrived. That is blocking, also called jamming. It is common, it is usually by design, and it has a different underlying cause than true ghosting.
You can see which of the two you have in about ten seconds with the interactive keyboard ghosting test: hold a combination down and the page shows every key your browser received as a separate event, so a missing key and an invented key look completely different on screen.
The keyboard matrix, and why it exists
A full-size keyboard has over a hundred keys, and giving each one its own pin on the controller chip would be expensive and physically awkward. Instead, keyboards use a matrix: the switches sit at the intersections of a grid of row and column wires, and pressing a key connects its row to its column. A grid of 8 rows by 18 columns covers 144 positions using 26 pins instead of 144.
The controller scans that grid constantly — energise row 1, read which columns respond, move to row 2, and so on, cycling through the whole grid hundreds or thousands of times per second. A column that reads back as connected during a given row's turn means the key at that intersection is down. Elegant and cheap, with one structural weakness.
Why three keys can create an ambiguous state
Suppose three keys are held down that form three corners of a rectangle in the matrix:
| Column 1 | Column 2 | |
|---|---|---|
| Row 1 | Key A — held | Key B — held |
| Row 2 | Key C — held | Key D — not touched |
With A, B and C held, current entering row 1 reaches column 1 through A and column 2 through B. But column 1 is also connected to row 2 through C. So a signal placed on row 2 can travel up column 1 through C, across row 1, and back down column 2 through B — arriving at exactly the point the controller would read if key D were pressed.
Electrically, the state "A, B and C are down" is indistinguishable from the state "A, B, C and D are all down". The controller has no way to tell them apart from the matrix alone. That fourth, uninvited key is the ghost, and the shape of the problem is why it is sometimes described as a sneak path or a phantom key.
What diodes do about it
The fix is a one-way valve. A diode is a component that lets current flow in one direction and blocks it in the other. Put a diode in series with each switch, all facing the same way, and the return path through key C is broken: current can go from the row to the column through a pressed switch, but it cannot come back the other way to complete the loop.
With a diode at every key, no combination of held keys can fake a key that is not down. That is the hardware foundation of what marketing calls anti-ghosting, and it is also the requirement for full n-key rollover.
Diodes cost money and board space, so many keyboards fit them only where they matter most — a "gaming cluster" of 20 to 30 keys around WASD, the modifiers and the space bar. Such a board is genuinely anti-ghosting in that region and behaves like a plain matrix everywhere else. This is why a keyboard can pass a test on one set of keys and fail on another.
Ghosting versus blocking: the distinction that actually matters
Keyboard designers worked out decades ago that a phantom keystroke is much worse than a missing one. A phantom S in a password field or a command line is a real problem; a fourth key that quietly does nothing is an annoyance. So controllers without full diode coverage are usually programmed to detect the ambiguous condition and refuse to report anything new until one of the offending keys is released.
| True ghosting | Blocking (jamming) | |
|---|---|---|
| What happens | A key you never pressed registers | A key you did press does not register |
| Cause | Ambiguous matrix state that the firmware reports rather than suppresses | The firmware detecting that ambiguity and deliberately dropping the input |
| How common today | Rare on keyboards made in the last fifteen years | Very common, especially on laptop and office keyboards |
| What you notice | Extra characters appearing in text | A character stops firing, or a movement key stops responding mid-game |
| Fix | Different keyboard; this is a firmware or design fault | Remap the combination, or use a keyboard with more diode coverage |
So when someone says a keyboard "has ghosting problems", the accurate description is almost always that it blocks certain combinations. The word stuck because the underlying matrix limitation is the same; only the firmware's response to it differs.
How to work out which one you have
- Open the ghosting test and click into the test area so it has keyboard focus.
- Hold down a combination you suspect — three or four keys that you use together, held simultaneously rather than tapped.
- Count the keys highlighted on screen against the keys under your fingers.
- Fewer keys shown than pressed means blocking. A key highlighted that no finger is on means true ghosting.
- Repeat on a different part of the board. Blocking is usually specific to particular row and column pairs, so a combination that fails on one side of the keyboard often works on the other.
If you want to know the maximum number of keys the board can carry at once rather than which specific combinations work, that is a rollover question — use the n-key rollover test instead, and see ghosting vs key rollover for why those are two different measurements.
What a browser test can and cannot see
A web page sees keyboard events after the keyboard controller, the USB or Bluetooth connection, and your operating system have all had their turn. It reports what arrived, not why something did not. That is still useful — the practical question "does this combination work on my machine?" is answered exactly — but a missing key could have been dropped by the matrix, by the connection, by the operating system, or by the browser itself.
Two other things routinely masquerade as a matrix problem. Some combinations are intercepted by the operating system or the browser before a page ever sees them, which looks identical to blocking on screen. And the six-key limit imposed by the USB HID boot protocol is a reporting limit, not a matrix limit — it caps how many non-modifier keys can appear in a single report regardless of how good the diodes are. That one is covered in the NKRO guide.
What this cannot tell you
No browser-based test can inspect your keyboard's matrix layout, count its diodes, or report hardware latency. It observes which key events reached the page, which is evidence about behaviour, not about internal construction. Every claim on this site is scoped to what the browser actually received — see the measurement methodology.
Does ghosting matter for you?
For typing, almost never. Ordinary typing rolls one key into the next and rarely holds more than two or three keys at once — the combinations every keyboard is designed to survive. It matters when software asks you to hold several keys simultaneously: a movement direction plus a sprint modifier plus an action key in a game, or chorded input in music and stenography software.
If a specific combination fails and you cannot change keyboards, remapping is the cheapest fix. Moving one key of the failing trio to a different row or column usually breaks the rectangle that caused the ambiguity. To measure input timing rather than key coverage, use the browser keyboard latency test, which reports median latency, jitter and consistency for your setup.
Frequently asked questions
Is keyboard ghosting the same as keys not registering?
No. Ghosting is an extra key appearing that you never pressed. A key that fails to appear when you did press it is blocking, sometimes called jamming. Blocking is far more common on modern keyboards, and the two problems have different causes.
Can a membrane keyboard be anti-ghosting?
Yes. Anti-ghosting does not require mechanical switches. Many membrane and rubber-dome keyboards add diodes or extra matrix traces to a limited set of keys, typically the gaming cluster around WASD, so those particular combinations register reliably while the rest of the board still blocks.
How many keys can a keyboard register at once?
It depends on the keyboard and the combination. Boards without per-key diodes may handle only two or three arbitrary keys before blocking. Boards with per-key diodes can in principle track every key, though the connection and the operating system may still limit how many are reported at once.
Does a browser ghosting test prove my keyboard has anti-ghosting?
It shows which combinations your browser received as separate key events, which is strong practical evidence. It cannot inspect the matrix or the diodes directly, and a failure could come from the keyboard, the connection, the operating system or the browser rather than the matrix alone.