Ookla Speedtest New Zealand fixed broadband 2022 Q3

Forums › New Zealand Broadband › Ookla Speedtest New Zealand fixed broadband 2022 Q3

1 | 2

cokemaster

Exited
4929 posts

Uber Geek

Retired Mod

Trusted

Lifetime subscriber

#2984523 19-Oct-2022 11:11

Big difference between using Speedtest.net for troubleshooting (irrespective of wifi-cpe vs ISP testing) vs using it to aggregate performance provider across a diverse set of access technologies and speed profiles.

webhosting

Loose lips may sink ships - Be smart - Don't post internal/commercially sensitive or confidential information!

allan

2044 posts

Uber Geek

ID Verified

Lifetime subscriber

#2984656 19-Oct-2022 12:53

michaelmurfy:

cokemaster: I concur with @freitasm. I bet that this is going to be used in marketing (and it has in the past)… so at least leaving some discussions online qualifying this is a “good thing” TM.

And bingo. They're running with it:

What's a "forth" year?

insane

3240 posts

Uber Geek

ID Verified

Trusted

#2984827 19-Oct-2022 18:26

Freudian slip perhaps? They envisage being top (by this metric) in the forthcoming year too?

insane

3240 posts

Uber Geek

ID Verified

Trusted

#2984829 19-Oct-2022 18:30

Talkiet:
Ookla are the default standard and once you remove the local issues (wifi, rubbish browser, outdated hardware, uploading to iCloud at the same time etc), it's an easy test for end users to run and gives them a number that they understand.

It's entirely possible to endorse their individual test (with a few caveats) procedures yet maintain their "Speedtest awards" are fundamentally flawed and in several provable cases are actively misleading. That's where I sit.

To me at least, it seems like the ISPs with the genuinely best performance don't actually need to go around using dodgy metrics to try and prove it.

N

Cant fault anything you said there. Can you update your reverse DNS records, segment connection type to IP ranges, or setup an individual AS per connection type to try level the bias?

Talkiet

4793 posts

Uber Geek

Trusted

#2984843 19-Oct-2022 19:17

insane:
Talkiet:

[snip]

Cant fault anything you said there. Can you update your reverse DNS records, segment connection type to IP ranges, or setup an individual AS per connection type to try level the bias?

While that's a list of things that could be done, having engaged with Ookla before (multiple times) on the validity of their methodology for the Speedtest awards, I don't believe that it would do any good.

There are ways to game the results and I can categorically state we've never even considered doing them.

Cheers - N

Please note all comments are from my own brain and don't necessarily represent the position or opinions of my employer, previous employers, colleagues, friends or pets.

mdav056

607 posts

Ultimate Geek

Subscriber

#2985358 20-Oct-2022 15:41

I initially thought that the 95% CIs would tell me something, and they indeed do -- the number in the sample!

gml

1 | 2