A question about segwit, blocksize and transaction fees

[ROT13]

Hypothetical Question-

I am mining bitcoin in 2016. Blocks are almost constantly full, people are quite often paying fees of 20 cents and up to ensure their transactions gets into a block promptly, with that number only looking to increase in the immediate future.

Why would I want to support segwit or a hardfork for an increased blocksize?

[franky1]

going with just segwit.

if 100% of peoples TX's were made using segwit. tx fee's of 20cents each ($500 @~2500tx) would go from ~$500 to ~$300
but
thats assuming 100% of tx are segwit.

then depending on how much pools open up their preferential settings within the new buffers. means more tx, which means more then ~$300 to offset the discount, you would see ~4500tx instead of ~2500

but thats assuming pools open up the preferential setting to fully use up the new allowances

ill try to explain

some think that all transaction data will be hit by a 4x discount. when in actual fact its just the signatures receiving the discount
some think that more space = less demand/less average fee, again wrong. pools wont jump straight to the max allowed buffer.

they would test the waters by opening their own safe preference allowance within the new buffer rules.
EG when the rule was 1mb for years, pools selected lower preference level over time to acclimatize themselves.
even in version 0.13.x of core there is a preference defaulting still at 0.7mb even though pools now raise it to 0.99x mb.
with segwit or even dynamic blocks. pools will try lets say 1.001mb weight first to test the water and then grow when they see the orphan risk is not causing issues.. they wont jump straight to max buffer rule on day one.

as for how much bonus(combined tx fee) will be, lets say 0.5btc combined was the old expectation / 20cents per tx old expectation.
you wont see a drop purely based on less demand. because pools wont open up the buffer fully.
so the 20cent per tx will still be expectant for legacy tx's, but if using segwit. only part of a tx gets a discount. which works out at a end result of paying 12.5cents total/tx instead of 20cents, which by allowing more transactions through over time offsets that discount abit, because the combined total will grow.

so because pools wont be jumping to max buffer instantly and users wont be jumping to 100% segwit usage instantly. dont expect much to change.

we are still going to see the fee war rising the price faster than the speed of users swapping over to segwit and pools to expanding their preferencial buffer.