In 1950, the doubling time of medical knowledge was estimated at 50 years. By 2010 it was 3.5 years. The projection for 2020 was 73 days — a projection whose horizon has already passed.1 Volume is only half of it. What is known does not merely accumulate; it turns over, and a meaningful share of what is taught as settled will be reversed within the career of the person taught it.2 A trainee who memorizes an answer this month may be carrying an expired one by the end of the rotation. 

Meanwhile, every one of us walks the corridor holding a device that opens onto more medicine than the libraries our teachers trained in. With a phone in hand, are we not all polymaths? The last individual who could plausibly hold the sum of human knowledge in one head died generations ago. Every resident can now reach all of it in seconds. There are no polymaths left, only people who know what to look for. So it is worth asking, because the trainees are asking it already. Do we need memory, or do we need a stable internet connection? 

Here is what the question leaves out. Every search has to be started by someone. Before anyone types a word, a clinician has already noticed that something is wrong. That is the one part nobody can look up. 

The operating room makes this visible. It is among the most densely monitored environments in medicine, and every value on the display is available instantly. Not one of them says what is happening. A rising end-tidal carbon dioxide, a climbing heart rate, a rigid jaw. Two of those are on the screen. The third is only available to someone standing close enough to feel it. Reading them together as malignant hyperthermia rather than hypoventilation is not a lookup. No monitor performs it, and no search engine is asked to. 

The pattern extends well beyond the operating room. Across 100 cases of serious diagnostic error, the most common contributing factor was faulty synthesis, most often premature closure. Inadequate knowledge was an uncommon cause.3 Those physicians were not short of information, they were short of the suspicion that would have sent them looking for it. The eyes cannot see what the mind doesn’t know. 

Which raises an uncomfortable question about how we decide who is any good. If retrieval is universally available, then recall is no longer what separates one clinician from another. Yet the instrument we still use to rank them measures precisely that. A systematic review and meta-analysis of USMLE Step 2 Clinical Knowledge pooled 43 studies. It found a moderate positive correlation with in-training examination scores (r = 0.52, 95% CI 0.45–0.59) and a very weak positive correlation with subjective measures of how residents actually performed (r = 0.19, 95% CI 0.13–0.25).4 The examination predicts the next examination.