[Chicago] Template Document

[Carl Karsten]

On Tue, Nov 2, 2010 at 1:33 PM, sheila miguez <shekay at pobox.com> wrote:
> On Tue, Nov 2, 2010 at 12:09 PM, Samir Faci <samir at esamir.com> wrote:
>
>> 1. Question 1
>
>> I could probably write a very simple one on my own.. just wondering if
>> there was any pre-existing technology / standard that I should look
>> at.
>
> I don't know of a standard or of a tool that does this for you, so I
> think you have to write it yourself. If you work at a large company,
> other people are probably annoyed and may have written things to do
> this too, so you should ask on internal back channels to see if anyone
> has. you might get lucky.
>
> IMO
> if you have to do it yourself, don't get too formal about what goes in
> there. I'd be very lazy and go as unstructured as possible and use
> something like <your favorite thing that is essentially a dictionary>.
> post the data somewhere via <your favorite protocol>, then when you
> need the report, GET it and yay.
>
> maybe you don't have to use identica or something, maybe you could use
> something that talks jabber over an internal jabber network? or
> whatever. some people might want the traffic to be only internal.
>
> anyway...
> I'm handwaving since I never was lazy enough to actually go through
> the effort to design something.
>
> if you find out something exists and is sharable, please point me to it.
>
>

This may be the next big thing.

The 'problem' with time tracking systems is they want to track time.
But trying to supply the data needed to do what it wants is so
annoying that the system fails.

What we need is a system that tracks activity, and if some pin head
wants that expressed as time, some simple math and the fantasy is
fulfilled.

What is activity?  all kinds of stuff: tweets, emails, code commits,
unit tests being run, phone calls, IMs...

I see a server that logs activity.  I see a bunch of clients that feed
it activity.  the trick is to make the clients as unobtrusive as
possible. personally, don't want them bothering me.   The clients spew
stuff at the server, and the server figures out how to classify it,
including unclassified.

Some clients will give it easy to classify stuff, like a code commit
spew can start with project:<repo_name> and the number of
lines/chars/files.

for extra credit: The server can use some machine learning foo and use
previously manually classified stuff as a pattern for new stuff.  like
if the same phone number comes up a few times, assume that all calls
to/from that number are part of the same project.

Oh yeah, back on my email to time report magic: I would review and
fluff it a bit - if some project seemed like it was getting more/less
than it's fair share, tweak the numbers.  but the tweaks were done at
the 'amount of stuff' level, so at the end of the day the math was
still 8h/%stuff   so I could boost one things stuff and not have to
figure out where to pull from.    I was salaried, so the fact that I
may have worked 10 hours but only logged 8 wasn't hurting me.  I would
be surprised if it hurt anyone.





-- 
Carl K