It was activated as a soft fork according to Bitcoin Improvement Proposal 148 (BIP 148), which enforced SegWit to be activated.
SegWit BIPs are 141 and 143 and it got activated like any other soft-fork like BIP-9.
BIP148 is actually more of a malicious attack on Bitcoin's principles. Nobody should not ever be allowed to "force" activation of any change without reaching majority's support. An important part that is missing from that BIP which is why I categorize it as a malicious attack.
SegWit transactions are simply smaller in size compared to legacy transactions because all witness data (signatures) is separated from transaction data. As a result, we get a lower fee per transaction vbyte.
I can't believe these two mistakes are still being repeated 7 years after activation of SegWit!!
SegWit transactions have lower weight compared to legacy, they are not smaller in size.
Also the witness data is NOT "separated" from transaction data. It is still a part of it like any other part like version, locktime, etc. It is just placed in a new location before the locktime in the serialization of transactions.
The reason why you end up paying a lower fee despite having bigger raw byte size is because fee calculation changed to be based on weight and virtual size.
By separating all witness data, SegWit effectively decreases size limits for each transaction and effectively increases Bitcoin’s block size limit, which helps to process more transactions per block.
Same mistake again.
The witness is still part of the transaction but because older nodes see SegWit output scripts as a script that anyone can spend, they don't need to see the signature. So it can be stripped
only when the block is being sent to the old nodes which no longer fully verify everything.
This way the capacity is increased in a backward compatible way.
You see the problem with saying "witness is separated" is that next someone is going to ask "where does it go" and "why can't I spend anyone else's coins if signatures are separated from transactions".