thats an expectation with/near 100% segwit usage
thats an expectation with/near LEAN tx..

the reality is not everyone will do lean TX's
the reality is not everyone will do segwit TX's

so your overselling expectation instead of thinking of rational realistic

put it this way.. 2009-2016.. how many blocks actually achieved the expectation of 7tx/s

oh and here is another nugget. although bitcoin never even got to achieve 7tx/s (4500) in 7 years in any reasonable amount of blocks

the maths of compressed keys and lean tx was a 10tx MAX.. but like i said even 7tx/s(undersold) was a push to even dare try coming close to and never really seen as any real longterm actual utility.
https://en.bitcoin.it/wiki/Maximum_transaction_rate

so your 14tx/s example is not a realistic longterm actual utility. but an extreme theoretical based on 100% this 100% that.
however
the realistic is that segwits 4500(7tx/s) expectation is still pushing the boundary's of what we will actually get. but atleast its not overselling it like your trying.

in short. its far easier to say
"since 2009. we may actually see 4500 regularly after 2016 due to segwit, which we expected to get 7 years ago"

sidenote:
im still laughing at lauda's magic block
Inputs 9,124
Outputs 9,157
Transactions 8,885

how many times does lauda realistically thing we are going to see blocks with 97% of users not only using segwit. but also doing 1 in 1 out tx