Course contentsShow
← TypeScript
Lesson 906 of 1,79621. The unknown TypeFree lesson

Type Safety Trade-offs

Explore the spectrum from strict (unknown) to permissive (any) and understand when each level is appropriate.

Type Safety Trade-offs

What you'll learn: How to navigate the spectrum from TypeScript's strictest (unknown) to most permissive (any) type, and when each level of strictness makes sense.

The Type Safety Spectrum

TypeScript offers you a sliding scale of type safety. At one extreme, unknown forces you to prove what you have before using it—maximum safety, maximum effort. At the other extreme, any lets you do anything—zero safety, zero effort. Between them lies a range of choices you'll make dozens of times per day.

Understanding the Trade-off

Think of type safety like airport security checkpoints:

  • unknown is like comprehensive screening: safe but time-consuming. You must verify everything before proceeding.
  • any is like no checkpoint at all: fast but risky. Anything gets through.
  • Middle ground (type guards, assertions, specific types) is like trusted traveler programs: balanced security with reasonable convenience.

When to Choose Strictness

Use unknown when:

  • Parsing external data (API responses, user input, JSON)
  • You genuinely don't know what type you'll receive
  • Safety matters more than convenience

Use any when:

  • Migrating JavaScript to TypeScript incrementally
  • Prototyping where types will be added later
  • Interfacing with truly dynamic code you can't model
  • The complexity of typing exceeds the benefit

Use specific types whenever possible—they're the sweet spot of safety and productivity.

The Real-World Balance

Most professional TypeScript codebases aren't purely strict. They use unknown at boundaries (inputs, parsing), specific types throughout business logic, and any in small, documented escape hatches. The goal isn't zero any—it's intentional any only where justified.

Key Takeaway: TypeScript's type safety is a spectrum, not binary. Choose unknown for untrusted inputs, any for justified escape hatches, and specific types everywhere else—let your use case, not ideology, guide you.