Arguments about dubbed and subtitled anime usually treat dubbing as a degraded copy. It is closer to a second performance built under constraints the original never had.
The animation was drawn to Japanese dialogue
In most anime production the voice recording happens after the animation, and mouth movements are drawn to a script rather than to a recorded performance.
Those movements are simplified, often cycling through a small set of shapes, which gives a dub more freedom than a live-action dub has.
Freedom is not the same as absence of constraint. The mouth still opens and closes for a set duration in a set rhythm.
An English line has to start when the mouth starts and finish when it stops.
Everything about the adaptation flows from fitting meaning into that window.
Syllable count is the real limit
Japanese and English package information differently, and a phrase that takes four syllables in one can take nine in the other.
The adapter's task is to find a line that carries the same meaning in the available space and still sounds like speech.
Where the meaning does not fit, something is cut, and what is cut is usually specificity.
Where the English is too short, material is added, which is why dubs sometimes contain lines that do not exist in the original.
Neither is carelessness. Both are consequences of matching duration.
Register is stated in English and implied in Japanese
Japanese encodes relationships through levels of politeness, honorifics and pronoun choice, often without saying anything explicit.
English has no direct equivalent. A character's formality has to be rendered through vocabulary, contractions and sentence construction.
This forces interpretive decisions. Deciding whether a character sounds stiff, warm or arrogant in English is deciding who that character is.
Two competent adaptations can arrive at different but defensible answers.
A viewer comparing them is often comparing interpretations rather than accuracy.
Casting a voice is casting a reading of the role
Japanese voice acting has its own conventions, including pitch ranges and delivery styles that do not map onto English-language expectations.
A young male character voiced by a woman is standard in Japanese production and unusual in English casting.
Directors have to decide whether to reproduce the original vocal texture or to find an English equivalent that produces the same impression.
Choosing equivalence rather than imitation usually produces a better dub and a larger apparent difference.
The character a viewer meets is shaped by that decision before a single line is recorded.
Jokes have to be rebuilt rather than translated
Wordplay does not survive translation. A pun that depends on two Japanese readings of a character has no English counterpart.
Adapters either replace the joke with a different one that lands in the same place, or let it fail and add a note that a dub cannot include.
Cultural references present the same problem, since a reference the original audience recognises instantly may mean nothing elsewhere.
The choice between preserving the reference and preserving the effect is the oldest argument in localisation.
Dubs generally choose effect, because a viewer who has to pause and look something up has already left the scene.
The mix is rebuilt, not just the dialogue
Dubbing requires a version of the soundtrack with the dialogue removed, and the new voices are mixed into that.
The mixing is done in a different studio with different equipment and different conventions about how prominent dialogue should be.
English mixes often place voices further forward than the original, which changes the balance against music and effects.
Some effects tied to the original dialogue, including breaths and vocalisations, have to be re-recorded.
The result can feel drier or fuller than the original for reasons that have nothing to do with acting.
Simultaneous dubs work under different pressure
Dubs produced alongside the original broadcast operate on a weekly turnaround, with scripts adapted and recorded within days.
Actors record without knowing where the story is going, which limits how much they can shape a performance across a season.
A dub produced after a series has finished can be cast and directed with the whole arc known.
That difference shows most in characters whose nature is revealed late.
Comparing a fast turnaround dub with a considered one is comparing two production schedules.
Subtitles solve a different problem, not the same one
Subtitles have their own constraint, which is reading speed. A line must be short enough to read before it disappears.
That produces compression as aggressive as anything in a dub, with subordinate clauses and qualifiers removed.
Subtitles also compete with the image, and a viewer reading text is not watching the animation in that moment.
Neither format delivers the original unchanged, because no format can.
The honest comparison is between two sets of trade-offs, chosen according to what a particular viewer would rather lose.