Every written language drifts from the speech it records, and Tibetan has drifted unusually far. Consonant clusters that the spelling still writes have been lost from the standard spoken language, voicing has been traded for tone, and syllables written with four or five letters are pronounced with two or three sounds.
It is tempting to read this as decay, and that reading is wrong in two ways. First, the spelling is not arbitrary: the silent letters are doing work, distinguishing words that would otherwise be identical and determining the tone the syllable is spoken on. A reader who knows the spelling can predict the pronunciation with fair reliability, which is not true of English. Second, the sounds the spelling records have not vanished everywhere. Dialects of the eastern regions, particularly in Amdo, retain clusters that Lhasa speech has lost, so the writing is closer to some living Tibetan than to others.
The comparison usually drawn is with English or French orthography, and it is only partly apt. Those systems fossilised largely by accident, through printing and inertia. Tibetan spelling was standardised deliberately, in periods of organised reform undertaken to make Buddhist translation consistent, so what looks like historical residue is in part the outcome of decisions taken about how the language should be written down.
The gap is real. It is not an accident.