You might be curious as to what kind of thing a promise is. But that of course would be to grossly misunderstand promises. A promise is not a thing but a claim, specifically a claim about a claim regarding a person’s future behavior. The promise says that this claim about future behavior is assured to be true, so that while we normally assume that any predictions about the future are unreliable we shouldn’t in this case. The promise then serves as a kind of meta-linguistic tool, which can be used to overcome certain dilemmas involving collective action.
Many have said that you shouldn’t break your promises because it is ethically wrong. I’m not exactly sure how sound that claim is; certainly to mislead people intentionally by making a promise known in advance to be false is probably unethical, but that is true for intentionally misleading people in general. And if the promise is broken without a malicious intent then all it seems to do is make us bad predictors of our own behavior, not unethical. For example, if I incorrectly predict tomorrow’s weather then I am not a bad person. Of course a promise might be treated as more than just a claim, we could choose to see it as an implicit contract between two people. But again, unless we accord special privileges to contracts, it doesn’t seem like breaking them is necessarily immoral. Which is not do deny that we have strong intuitions against breaking promises, but these intuitions are not because breaking promises is immoral, but because there are strong practical reasons not to break them (or so I claim).
The practical reason not to break promises is that breaking a promise renders you unable to make more promises. Of course you can still say that you promise, but the people who you are promising to will treat your promise just as a simple prediction of your future behavior, and not one that is especially certain. The reason this is so is because communication rests fundamentally on reliability. A listener takes the speakers words to mean what they do in part because the listener thinks the speaker can reliably use the terms to designate what they are supposed to designate. By breaking a promise you thus indicate that you are unreliable at using “promise” to designate claims about future behavior that are certain, and hence listeners no longer take your use of that word as designating anything at all.
Of course this is not a phenomenon that is unique to promising, it could happen, in theory, with any piece of language. For example, someone could be unreliable at using color words, meaning that they randomly match color words with actual colors. Such a person would thus be effectively unable to communicate about color, because calling the color of something “red” would not be taken by the listener to indicate anything at all. The use of ethical terms is similar. It is quite possible to use ethical terms just to designate what you like and don’t like, and not what is really right or wrong. But if you do that then people simply stop taking your use of them as indicating right and wrong, and only as indicating your preferences. Communication, as I have noted above, is based on trust, the listener trusts the speaker to use words in their usual senses. When that trust is broken communication can’t happen, and that is to the disadvantage of both the listener and the speaker. Of course promises do stand out as being a situation where it is easy for listeners to lose trust in the speaker. It is quite possible to misjudge color on occasion and still be able to communicate about color, but breaking only a few promises leads people to ignore future occurrences of promising. I don’t think, however, that this is because promising is something special, rather it is just that there are fewer mitigating factors. When I misjudge color there are many possible reasons for that, not all of which involve me being a bad judge of color. Thus people may still believe me to be a generally reliable judge of color in spite of a few errors. But since promises are just about our future actions there are far fewer mitigating factors to appeal to; since we are in control of our actions it is hard to explain why we didn’t act as promised (and since we are only supposed to promise when we can be sure, the fact that something made us unable to keep our promise indicates that we couldn’t be as sure as we thought, which indicates that we are bad at judging when we can make promises). Thus even a few broken promises mark a person as an unreliable promiser.
And not only does breaking promises make it nearly impossible to promise in the future (or at least to get people to take your promises as reliable), but it makes people question your motives for promising in the future as well. As mentioned above since we control our actions any promise breaking appears intentional. And if it is intentional then people will suspect that you made the promise in order to achieve some benefit for yourself. Which means that in the future making a promise will lead people to conclude not that you will actually act in a certain way, but that you want them to believe that you will act that way. And such reasoning leads to a general suspicion about motives and about how self-serving you are. Which in turn leads to mistrust, since every action can appear self-serving upon close scrutiny. So the moral here is to keep your promises. Or, if you can’t do that, don’t be caught breaking them. Or, if you can’t do that, have at least an excuse ready that doesn’t involve selfish reasons for breaking the promise.