Halakhah does not assume that anything older is necessarily more precise despite the principle yeridat hadorot (ירידת הדורות), and this is scientifically not reasonable either. Assume you write an important letter, and you ask a little child living next door to copy it, and a week/year/decade/century later you ask a professional copy editor to copy your same letter. Which copy would you trust more?
Instead, in the Yerushalmi Taanit 4:2 it is taught:
שְׁלֹֹשָה סְפָרִים מָצְאוּ בָעֲזָרָה. סֵפֶר ״מְעוֹנֵי״, וְסֵפֶר
״זַעֲטוּטֵי״ וְסֵפֶר ״הִיא״. בְּאֶחָד מָצְאוּ כָתוּב מְעוֹן אֱלֹהֵי
קֶדֶם, וּבִשְׁנַיִם כָּתוּב מְעוֹנָה אֱלֹהֵי קֶדֶם. וְקִייְמוּ
שְׁנַיִם וּבִיטְלוּ אֶחָד. בְּאֶחָד מָֽצְאוּ כָתוּב וַיִּשְׁלַח אֶת
זַעֲטוּטֵי בְּנֵי יִשְׂרָאֵל, וּבִשְׁנַיִם כָּתוּב וַיִּשְׁלַח
אֶת־נַעֲרֵי בְּנֵי יִשְׂרָאֵל. וְקִייְמוּ שְׁנַיִם וּבִיטְלוּ
אֶחָד. בְּאֶחָד מָצְאוּ כָתוּב תֵּשַׁע הִיא, וּבִשְׁנַיִם כָּתוּב
אַחַד עֶשְׂרֵה הִיא, וְקִייְמוּ שְׁנַיִם וּבִיטְלוּ אֶחָד.
Three scrolls were found in the Temple courtyard, the Meone scroll, the Zaatute scroll, the hi scroll. In one they found written meôn (מְעוֹן) is the preexisting God, but in two was written meônāh (מְעוֹנָה) is the preexisting God (Moses V. 33:27). They confirmed the two and annulled the one. In one they found written he sent the zaatûtē (זַעֲטוּטֵי) of the Children of Israel but in two was written he sent the naarē (נַֽעֲרֵי) of the Children of Israel (Moses II. 24:5). They confirmed the two and annulled the one. In one they found written nine היא, but in two they found eleven היא. They confirmed the two and annulled the one.
The same is discussed in Sifrei Devarim 356 and in Soferim 6. Therefore, our sages of blessed memory used the majority principle to decide which manuscript was correct if they had conflicting versions. Despite not being flawless,* this is a scientifically valid method as well, and this is how they reconstructed missing parts of the Aleppo Codex as well.
* The choice of the manuscripts for comparison is not trivial.