A critical question as regards the new coronavirus has been to what extent it is transmitted by asymptomatic people. But really there isn't just one definition of "asymptomatic". Instead, there are several possible definitions, each of which is useful in a different way:
Truly asymptomatic, where the infected person can't detect the slightest hint of anything being wrong, and neither can anyone else.
Asymptomatic as far as anyone else can tell, but where the victim can tell that there's something wrong. Fever, for instance, can often be concealed (though not from infrared thermometer checkpoints, such as have become common in Asia).
Having symptoms (obvious or concealable), but not ones that are thought of as being coronavirus symptoms.
That last definition is where things start getting really interesting. If you look at the CDC's "Symptoms of Coronavirus" web page today, it tells you that the symptoms to watch for are fever, cough, and shortness of breath. They're far from alone in that advice; many others are naming those three symptoms as the ones to watch for. One set of Chinese guidelines, for instance, states that:
Main manifestations include fever, fatigue and dry cough. Nasal congestion, runny nose, sore throat, myalgia and diarrhea are found in a few cases.
But the truth seems to be that, as one paper puts it,
COVID-19 can present as a mild upper respiratory tract illness. Active virus replication in the upper respiratory tract puts the prospects of COVID-19 containment in perspective.
It sure does; in blunter language, the coronavirus can look like a common cold and spread like a common cold. This means that people who just think they have a cold should be warned against spreading it -- and, ideally, tested for the coronavirus and treated with antivirals if positive. That's "ideally" because at the moment tests are slow and scarce, and the available antiviral treatments are all of questionable effectiveness. But antivirals almost always work best when started early, so it's an ideal that we should work towards.
As for why so many people would be wrong about such an important thing as the usual course of this disease, as Chris Masterjohn put it, commenting on that same paper:
Many cases of COVID-19 are identified based on symptoms first, and then testing to confirm. Since the testing is only done when the symptoms seem like a compelling case of COVID-19, the testing is biased toward the prevailing beliefs about the tell-tale symptoms. This makes the bias confirm itself: doctors only test when the symptoms cluster according to the early beliefs; therefore, cases where the symptoms differ from those beliefs are never tested and never confirmed.
Now, if the upper respiratory symptoms persisted throughout the course of the disease, they too normally would be added to the list of its symptoms. But the paper also found that as the disease progressed, their patients cleared the virus from the nose and throat even as it was still worsening in the lungs. People usually don't go to the doctor for a mere cold, and by the time they do go to the doctor, enough of their cold-like symptoms have faded that those symptoms aren't thought characteristic of the disease. A good old-fashioned doctor, of the sort who actually spent considerable time listening to patients, wouldn't make that mistake; but good old-fashioned doctoring has been buried under heaps of bureaucracy.
Yet when trying to control an epidemic disease, recognizing its first symptoms is crucial. Shunning everyone who seems to have even a mild cold is an unpleasant thing to do, but it's still better than shunning them plus shunning people who have no symptoms whatsoever.