TL;DR

BIP39 was introduced two years after Electrum. BIP39 seeds include a checksum, in order to help users figure out typing errors. However, BIP39 suffers the same shortcomings as early Electrum seed phrases:

A fixed wordlist is still required. Following our recommendation, BIP39 authors decided to derive keys and addresses in a way that does not depend on the wordlist. However, BIP39 still requires the wordlist in order to compute its checksum, which is plainly inconsistent, and defeats the purpose of our recommendation. This problem is exacerbated by the fact that BIP39 proposes to create one wordlist per language. This threatens the portability of BIP39 seed phrases.

BIP39 seed phrases do not include a version number. This means that software should always know how to generate keys and addresses. BIP43 suggests that wallet software will try various existing derivation schemes within the BIP32 framework. This is extremely inefficient and rests on the assumption that future wallets will support all previously accepted derivation methods. If, in the future, a wallet developer decides not to implement a particular derivation method because it is deprecated, then the software will not be able to detect that the corresponding seed phrases are not supported, and it will return an empty wallet instead. This threatens users funds.

For these reasons, Electrum does not generate BIP39 seeds. Starting with version 2.0, Electrum uses the following Seed Version System, which addresses these issues.

Electrum 2.0 derives keys and addresses from a hash of the UTF8 normalized seed phrase with no dependency on a fixed wordlist. This means that the wordlist can differ between wallets while the seed remains portable, and that future wallet implementations will not need today’s wordlists in order to be able to decode the seeds created today. This reduces the cost of forward compatibility.