Re: #578: getting real-ish numbers for option 3

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 28/10/2014 3:48 p.m., Nicholas Hurley wrote:
> On Mon, Oct 27, 2014, at 19:10, Amos Jeffries wrote:
>> Huh?  -10% bandwidth reduction == "miniscule" savings ?
> 
> *sigh* I should've known someone would pull that out,
> wrong/misleading though it may be in this case. Taking 10% off of
> an already really small number leaves you with (drumroll,
> please)... an insignificantly smaller number!

Statistical significance is (N >= 0.05) .

Only in *absolute* terms is the number small. At these compression
scales it means the difference of a whole extra request per packet.
Scale that up to TB/s rates and call it insignificant!


> And let's not forget, that (miniscule) 10% was in only one case; 
> the other cases Willy showed were even less impressive than that
> (and I maintain that his "amazing" 10% is in no way impressive to
> begin with).
> 

What I see in Willys' numbers is a handful of use-cases. Some of which
are not benefiting much - but most importantly not being degraded),
and some which have been causing controversy getting a significant %
reduction.

I also see people arguing earlier over reduction of a few individual
bits on the wire. Here is several whole bytes easily shaved off in a
way that reduces controversy. What's not to like?

Amos

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (MingW32)

iQEcBAEBAgAGBQJUTx8IAAoJELJo5wb/XPRj9nUH/iWU2BVH2Ns5wv+9+YaD9VRe
/BykcGfSNhXAt8ZS6mlefaVYAq4z7EgxBK8PoMKIRGufm7GKz/iIcv9IQmFjj6Zq
T4eb9sFxqpZxFTQXeHXp8UIlum6RmWHAve7dm/4y27+uEawr/KS9gD32B1jzhkIq
B9QGammB2eMzVvsYNVjdN0+560DGGEDHcckC/VEzNBShaTBqdp8/wOR+f/aYMKhx
wqBUz3ikmyi0NQFHcWyqDJ6iteDswKiVAp6FMPaO75FrdsbqebfojDkWI2rzhfpg
n70y5OT8upwOfaNlTqlW2BlL4bTprbrMagRHcrGMwc4a+tun9GzmkRyr/7Q5h2M=
=bu/i
-----END PGP SIGNATURE-----

Received on Tuesday, 28 October 2014 04:44:28 UTC