What Is Garbage Collection in Java? — Reachability and Reference Strengths

Java's garbage collector automatically reclaims memory used by objects no longer reachable from the application. The key insight is that GC is reachability-based, not reference-count-based — it starts from a set of GC roots like thread stacks and static fields, and any object it cannot reach is garbage.

How GC decides what's garbage:

• Walks references from GC roots: thread stacks, static fields, JNI references, system class loader
• Anything unreachable is collected — including circular reference cycles, which reference counting cannot handle
• This is why Java does not need explicit free() calls and why circular references do not leak

Reference strengths from strongest to weakest:

• Strong — the default = assignment; pins objects in memory until they go out of scope
• Soft — collected only under memory pressure; designed for caches but often worse than bounded caches like Caffeine
• Weak — collected at the next GC; powers WeakHashMap and ThreadLocal cleanup
• Phantom — used for cleanup notification after finalization; backs the modern Cleaner API

Common misconception to address:

• GC does not prevent all memory leaks — it only collects unreachable objects
• The most common Java memory leaks are: static collections that grow forever, listener registrations that are never removed, and ThreadLocal values in pooled threads

In an interview, I'd emphasize that understanding reachability is what lets you debug a production heap dump — knowing the GC algorithm is secondary.