Quote from: tl121 on December 01, 2015, 08:23:50 PM

From a little watching of my connection activity, I observed that a lot of the upload bandwidth was loading older blocks. A limit on bandwidth provided for uploading older blocks would certainly help.

Implemented in Bitcoin Core git master a couple months ago, will be in 0.12; along with many other resource control tools. Though if widely used it will further congest getting blocks to begin with. It's a good trade-off, but not without costs. None of this is magic: cutting down this source of traffic or that is mostly just constant factor reductions. To operate the system needs a safety margin, and we're bumping up against actual limits.

Quote from: Carlton Banks on December 01, 2015, 08:29:28 PM

Not sure what happened with that, looks like it didn't make it into 0.12, but perhaps I'm not looking hard enough. Or maybe there was a good reason not to include it (apparently network message anti-spamming measure did make it into 0.12)

You weren't looking hard enough

, it's called -maxuploadtarget see also doc/reduce-traffic.md (though not all of the features are documented yet). (amusingly, this was one of the relatively few PR's that Mike Hearn commented on-- opposing it.)

Quote from: sgbett on December 01, 2015, 08:59:23 PM

That does not mean I think the crowd will choose big blocks, it means that I think if bigger blocks are needed then I think that something will happen to ensure that bigger blocks happen.

This was one of the big fallacies involved in the "crash landing" spin by Gavin and Hearn; this notion that Bitcoin would willfully commit suicide. Cranking the scale ahead of addressing scaliablity is/was controversial (not just among the most technical, though it's nearly universal there); but if it were strongly _necessary_, and better than the harm of not; then it would no longer be. That we're not there now, shows it isn't. QED. And there is a tone of activity gone on to improve scalablity at all levels of the system, from micro optimizations, to protocol design.