I love how after you were unable to refute my point about the entire block needing to validated, you then tried to switch arguments and started talking about upload vs download speed.
Nobody mentioned upload nor download speed. Since you insist on rebuttal, I have included it in this post. It was really trivial, and quite cute that you think you had a real *winning* argument.
![Cheesy](https://bitcointalk.org/Smileys/default/cheesy.gif)
Switchng arguments instead of conceding you were wrong is a clasic symptom of an arrogant know-it-all (or a troll).
Nobody is switching anything; these are growing signs of you delusional.
Don't try to change arguments again. You were against big blocks (32MB). The latency issues are for "small amounts" of data (1mb). Big blocks = medium-large amounts of data.
You have no real argument, and this is probably because you have a major in something useless (e.g. arts or history). They were all designed to reduce propagation times, which was (or depending who you ask, still is) an issue today (regardless whether you're talking about 1, 8 or 32 MB). Both of those originated from the Bitcoin Relay Network:
http://bitcoinrelaynetwork.org/The Bitcoin Relay Network .... b) decreases block propagation times between miners.
Bandwidth is obviously not negligible. However, the exact amount of constraint that is creates is inconsistent. You are forgetting certain things that factor into this:
1) Compact blocks, i.e. reconstruction of blocks from block sketches.
2) SPV mining (which is unfortunately still a thing)[1].
Now the worst case for bandwidth constraint (concerning orphan rates) is transacting a maximum sized block, containing transactions that no other mining pool/node has (or nobody is running a client supporting compact blocks). What is the likelihood of even a 1 MB block, from which no other party has any included transactions? Here's a somewhat old, down-voted by trolls and shills, video based on data:
https://www.youtube.com/watch?v=Y6kibPzbrIcNote: This is data prior to the (public) existence of compact blocks, FIBRE, et. al. I couldn't find any recent data on this.
[1] - These two words completely destroy your previous statement:
Orphan rates are primarily affected by bandwidth, not first-bit latency, because you need the entire block to be transmitted and then validated before miners will build on it or subsequently relayed.
They don't need to validate it before building upon it nor before relaying it. Learn what SPV mining is. Or as franky1 puts it, LEARN BITCOIN.