|
qirex posted:big data is funny because instead of using some intelligence to figure out what tiny fraction of your metrics are most important to your business just capture everything and Big Data will tell you a bunch of dumb correlations that will confuse your senior leadership for years! [please keep paying your support contract during those years] look if we don't record every mouse motion and mis-click cross-referenced against every revision of every page, how will we understand user behavior?
|
#
?
Nov 26, 2014 16:26
|
|
|
|
| # ? Apr 24, 2024 07:56 |
|
|
Notorious b.s.d. posted:i talked to a guy who was working on a migration from linux/x86 to aix/power because ibm said it would save the company money on db2 lmbo
|
|
#
?
Nov 26, 2014 17:33
|
|
|
make it soap, number 1
|
|
#
?
Nov 26, 2014 17:36
|
|
|
mishaq posted:lmbo lmbo
|
|
#
?
Nov 26, 2014 17:36
|
|
|
MALE SHOEGAZE posted:make it dubstep, number 1
|
|
#
?
Nov 26, 2014 17:37
|
|
|
drop the table
|
|
#
?
Nov 26, 2014 17:40
|
|
|
Notorious b.s.d. posted:i talked to a guy who was working on a migration from linux/x86 to aix/power because ibm said it would save the company money on db2 moving to power is definitely how you save money
|
|
#
?
Nov 26, 2014 21:27
|
|
|
Notorious b.s.d. posted:look if we don't record every mouse motion and mis-click cross-referenced against every revision of every page, how will we understand user behavior? users invariably behave like stunted idiots. there we go, I saved everyone so much time and effort.
|
|
#
?
Nov 26, 2014 21:50
|
|
|
BooLoo posted:users invariably behave like stunted idiots. there we go, I saved everyone so much time and effort. already learned this from valve game commentary but good to reinforce every once in a while
|
|
#
?
Nov 26, 2014 21:51
|
|
|
Notorious b.s.d. posted:look if we don't record every mouse motion and mis-click cross-referenced against every revision of every page, how will we understand user behavior? frequently this is necessary to show that some exec's brilliant idea is actually harmful to the business
|
|
#
?
Nov 26, 2014 21:53
|
|
|
carry on then posted:already learned this from valve game commentary but good to reinforce every once in a while sitting down next to and just watching a user do their thing on a piece of software you've built or helped build is a great way to end up realizing that we only ever made it out of the swamp and onto dry land by luck.
|
|
#
?
Nov 26, 2014 21:55
|
|
|
after years of watching doctors use our software i am never setting foot in a hospital again it should just know what you're allergic to, im not putting that in. that's clerical work i have other things to do
|
|
#
?
Nov 27, 2014 00:10
|
|
|
qntm posted:meanwhile in the past
|
|
#
?
Nov 27, 2014 00:22
|
|
|
lol like lotus enables anything at all except the opportunity to fix lotus
|
|
#
?
Nov 27, 2014 01:38
|
|
|
we did a firedrill yesterday in a production support environment by getting the DC to hard shutdown a websphere messaging gateway and it is supposed to be redundant but it turns out the standby node won't switch to active unless the master shuts down gracefully
|
|
#
?
Nov 27, 2014 02:56
|
|
|
gently caress IBM forever
|
|
#
?
Nov 27, 2014 02:57
|
|
|
throw a big-ip in front of it duh
|
|
#
?
Nov 27, 2014 03:00
|
|
|
expecting enterprise poo poo to fence correctly is naive at best
|
|
#
?
Nov 27, 2014 03:01
|
|
|
qirex posted:frequently this is necessary to show that some exec's brilliant idea is actually harmful to the business a/b testing only requires you to measure core metrics, not a gigabyte of data reflecting the pointer position
|
|
#
?
Nov 27, 2014 03:37
|
|
|
pram posted:throw a big-ip in front of it duh pram posted:expecting enterprise poo poo to fence correctly is naive at best
|
|
#
?
Nov 27, 2014 03:37
|
|
|
pram posted:expecting enterprise poo poo to fence correctly is naive at best this is depressingly correct
|
|
#
?
Nov 27, 2014 03:51
|
|
|
but on the other hand ughhhh
|
|
#
?
Nov 27, 2014 03:51
|
|
|
|
| # ? Apr 24, 2024 07:56 |
|
|
Notorious b.s.d. posted:a/b testing only requires you to measure core metrics, not a gigabyte of data reflecting the pointer position a/b testing requires you to know what the hell you want to know in the first place, takes place over a defined period of time, and is simple enough that it's usually pretty clear which thing works better big data is appealing to execs with no particular plans for the data, so they can collect loving everything forever and stash the data away until some manager wants to cherrypick a few data points that support their newest pet project, or until the company decides to hire a consultant or computer program to (pretend to) analyze all the data and make all sorts of pronouncements about what the company should do
|
|
#
?
Nov 27, 2014 16:34
|
|