Suppose we have three users: (AS) Andy Smith (BS) Andy Smith's brother Bobby Smith who can telepathically communicate with his computer (CS) An alien computer scientist

Because the process of programming involves the translation from natural language to machine language (logic), there will necessarily be various syntactical preferences depending on the natural language of the programmer and the input method.

For this example we will assume AS and BS have the same basic cognitive structure and natural language. The only difference is that BS is telepathically linked to his computer, whereas AS must use a keyboard. Presumably there now is no barrier for BS to use a purer translation of natural language to machine language. Because there is no time cost for typing, BS can represent lambda as 'lambda' instead of 'l' or simply parens. However, AS will likely choose abbreviations, since to them they represent an ease of use in the translation process from natural language to machine language. However, probably he will use only English language derived abbreviations of no less than three characters, as anything else would be outside his cognitive context.

Because our alien programmer (CS) thinks in a completely different language, the preferred syntax will also be different. Perhaps CS will want to represent concepts (such as lambda) as a hexadecimal number. This can hardly be said to be better or worse as machine logic does not change, only the representation/syntax changes.

So I think if I had to rewrite my earlier statement I would get rid of the word 'preferences,' which implies subjectivity. Rather, I would suggest there are different optimal syntaxes depending on the point of origin (or point of translation from natural language to machine language) but likely only one optimal representation as machine logic (per problem).