Blocking the line instead of removing the guard
Quick answer: Interference means sacrificially placing a piece on the line between an attacked piece and its defender, cutting the defence. The defender is still there, it just can't see through your piece any more. It differs from deflection, which lures the defender away entirely. The two classic problem themes are Novotny, where the sacrifice sits on the intersection of a bishop line and a rook line, and Plachutta, where both defenders move orthogonally.
Every club player learns to remove a defender. Capture it, chase it, or lure it away with a bigger threat.
Interference does something stranger. It leaves the defender exactly where it is, and makes it useless by standing in the way.
The first time you see one it looks like a blunder, because you're putting a piece somewhere it can be captured for apparently no reason. Then you notice that capturing it breaks the very line that was holding the position together, and it clicks.
The idea, stripped down
Say Black's rook on d8 is defending a knight on d5. You want that knight.
Deflection would attack the rook, or offer it something juicy elsewhere, so it leaves d8.
Interference just puts a piece on d6 or d7. The rook is still on d8, still theoretically defending down the d file, but now it's looking at your piece instead of at its knight.
Black's problem is that capturing the interposed piece often makes things worse, because the capture uses the very piece that was doing the defending, or opens a different line. That's the paradox at the heart of the tactic and it's what makes it satisfying when it works.
Three conditions have to hold:
The defender has to defend along a line, so a rook, bishop or queen. Knights cannot be interfered with, because they jump. Neither can a king or pawn defending an adjacent square. This narrows where you should even look.
The piece being defended has to be worth more than what you're sacrificing, or the arithmetic doesn't work.
And the interposition has to create a real threat, otherwise your opponent just takes it and thanks you.
The interference entry on Wikipedia) is unusually clear on the formal definition, and worth a read if the distinction from deflection still feels fuzzy.
I've only ever played one genuine interference in a serious game, and I found it by accident. I was looking for a check, noticed the square was defended by a rook down a file, and realised almost as an afterthought that a knight could sit in the middle of that file with tempo. My opponent took it and immediately lost the piece the rook had been guarding. It felt like cheating.
Interference versus deflection
These get confused constantly and the difference is worth nailing down, because it changes what you look for.
Deflection removes the defender from its post. You attack it, or you offer a capture it can't refuse, and it goes somewhere else. The defender moves.
Interference leaves the defender in place and cuts its line of sight. The defender stays put and stops working.
Decoy, which is the third member of this family, lures a piece to a specific bad square rather than simply away.
The practical consequence is what you scan for. If you're hunting deflection, you look for defenders you can attack or tempt. If you're hunting interference, you look for long lines and empty squares sitting on them.
That second scan is much rarer and it's why interference gets missed. Nobody looks at an empty square in the middle of a line and thinks "what if I put something there and let it be taken".
The existing breakdown of deflection as a tactic covers that side properly, and reading the two together is the fastest way to stop mixing them up. They're siblings, not synonyms.
Novotny and Plachutta
These two names come from composed problems rather than practical play, but knowing them sharpens what you're looking for.
A Novotny sacrifice lands on the square where two defensive lines cross, and crucially those lines belong to pieces that move differently, typically a bishop and a rook. Whichever piece captures your sacrifice, it blocks the other one's line in the process. The defender is forced to choose which defence to abandon, and there is no third option.
That's a beautiful bit of geometry. One square, two defences, and capturing either way breaks something.
The name comes from a problem composer, as most of these do, and the theme has a long history in the problem world where whole tournaments are built around a single idea. Practical players tend to be dismissive of that tradition, which I think is a mistake. Composed problems are where the pure form of a pattern gets isolated, and seeing the pure form is what makes the messy version recognisable later.
A Plachutta is the same trick where both defending pieces move along the same kind of line, so two rooks, or a rook and a queen operating orthogonally. Because they move alike, the capture overloads the capturing piece with two jobs it can't both do.
You will almost never construct one of these deliberately in a real game. What matters is the pattern recognition they train: the habit of looking at intersections of enemy lines and asking what happens if something lands there.
I'd honestly say this is the least practical tactic in the standard list, in terms of how often you'll play it. It's still worth an hour, because the scan it teaches, watching where enemy lines cross, catches other things too. Overloaded pieces, pins, and skewers all live in the same geometry, and the overloading guide covers the closest relative.
Training your eye for it
Since it's rare, the question is how much effort it deserves. My answer is: a little, aimed precisely.
Do not grind hundreds of interference puzzles. The return is poor because the pattern appears so infrequently in practical games that you'll forget it before you use it.
Instead, add one question to your existing calculation routine. When a defended piece is stopping you from playing something, ask: is that defence coming down a line, and is there an empty square on that line I could occupy with tempo?
Ten seconds. That's the whole training programme.
I've been running that single question for a while now and the honest result is that it fires maybe once every thirty games. But it costs nothing, and the games where it does fire tend to be the ones where a position looked completely blocked a moment earlier.
The other thing that helps is playing through composed problems occasionally rather than only tactical puzzles from games. Composed problems are where these themes live, and they stretch your calculation in ways that game positions don't, because they're built to have exactly one absurd solution.
There's a broader point here about how tactics study works. Learning names is not the goal. The goal is that when you look at a position, certain configurations produce a small itch that makes you calculate further. Names are just filing labels for the itch.
If you want the efficient version of building that instinct, volume on mixed puzzles beats depth on any single theme. The puzzle trainer mixes patterns rather than telling you the theme in advance, which is much closer to what happens over the board, and the 30 day calculation routine is a sensible structure to hang it on.