Online scheduling is a game against an adversary who reveals jobs one at a time. Each job arrives with a processing time. You must decide what to run now without knowing what's coming next. The competitive ratio — how much worse your schedule is compared to the optimal offline schedule that knows everything in advance — measures how much ignorance costs.
Full clairvoyance means knowing the entire future. It makes online scheduling trivial — just run the offline optimal. Zero clairvoyance is the standard online model — you know nothing about future arrivals. ε-clairvoyance is between: you get a small peek into the future. Specifically, when a job arrives with processing time p, you know the future for the next εp time units. Longer jobs give you more warning. Short jobs are still surprises.
The remarkable fact is that even tiny clairvoyance (ε close to zero) dramatically improves scheduling. The competitive ratio drops from polynomial (without clairvoyance) to constant (with any positive ε). The transition is discontinuous: zero versus any clairvoyance is the difference between a fundamentally hard problem and a fundamentally easy one.
The simpler analysis now makes this accessible. The previous proofs that ε-clairvoyance suffices for constant competitive ratio were technically intricate. The new proof is streamlined — the core argument is cleaner, the constants are better, and the technique is more modular. The result hasn't changed, but the explanation has become proportional to the insight: a small window into the future is enough.