A loss on paper gets filed as something that has not yet happened. The price could return, the position is still open, and the matter remains formally unsettled. A realised loss gets filed as something concluded, and what has been concluded extends past the sum of money to include confirmation that a judgement was wrong. Measured against net worth on any given day, the two states produce exactly the same number, and no accountant would treat them differently. The distinction exists only in the mind of the person holding the position.
Setting tax aside, the economic difference between them is close to nothing. Holding an asset that has fallen and holding the equivalent cash after selling place a person in the same financial position, with the same options available from that point onward. The whole of the difference sits in classification. One is filed as unfinished and the other as finished, and people handle those two categories in entirely different ways without ever having decided to. The filing happens automatically and is almost never revisited.
This explains why selling a losing position is so difficult. The difficulty does not come from an assessment of the asset's prospects. In most cases the holder's view of those prospects has already changed, and they would not establish the position today. It comes from what the act of selling does to the record. It converts an unsettled matter into a settled one, and the settlement covers the holder's own judgement along with the money. Continuing to hold leaves the question open, and an open question demands nothing from anybody.
The same mechanism explains a phenomenon that looks unrelated at first. Switching a losing position into a different asset is considerably easier than simply selling it and holding cash. The financial outcome can be identical in every respect that matters. A switch gets filed as a reallocation while a sale gets filed as taking a loss, and only one of those requires an admission. A great deal of unnecessary trading happens inside that gap, and the people doing it generally believe they are repositioning.
Mental accounting plays an unmistakable role here. Each position gets managed as a separate account with its own profit and loss status. The state of the portfolio as a whole receives comparatively little attention, despite being the thing that actually matters. Decisions consequently get made at the wrong level. The question that ought to be asked concerns whether the overall allocation still makes sense given what is now known. The question actually being asked is whether this one compartment can be closed at par.
One test bypasses the entire apparatus. Suppose the whole portfolio were converted to cash today with no tax and no transaction costs of any kind. Which positions would be bought back, and at what sizes. The answer usually differs from the existing portfolio in several places. The difference consists precisely of holdings sustained by classification effects, with no current view of their merits behind them. Those are the positions worth examining first, and they are also the ones least likely to be examined at all.
The force of the question comes from its removal of the existing categories. Cash has no profit and loss status and no compartments waiting to be closed. Every asset therefore has to earn its place again on its own terms, in competition with everything else available. Tax and transaction costs mean that acting fully on the answer is rarely sensible in practice. Knowing the answer and acting on it are separate things, and most people have never asked the question at all.