RE: [eclipse.org-eclipsecon-program-committee] Standard talks spreadshee

["Scott Rosenbaum" <scottr@xxxxxxxxxxxxxxxxxxxxx>]

Yes but who's on first?

scott

-----Original Message-----
From: eclipse.org-eclipsecon-program-committee-bounces@xxxxxxxxxxx
[mailto:eclipse.org-eclipsecon-program-committee-bounces@xxxxxxxxxxx] On
Behalf Of Oisin Hurley
Sent: Sunday, January 10, 2010 4:55 PM
To: Eclipsecon Program Committee list
Subject: [eclipse.org-eclipsecon-program-committee] Standard talks
spreadsheet - percentage allocations, numbers, etc.

I've create one more (final!!!) spreadsheet in the shared folder.
It is called 'Standard Talks - Percentages'.

It takes information from the 'updatable 2' spreadsheet, and from
the 'haggling-standard-talks spreadsheet' and attempts to bang
the two together to provide information that will be useful in
getting us to a final program.

It needs some explanation, because you'll be wondering wtf
is going on. First I better write what I am trying to illustrate
here, but even before that I need to state that the spreadsheet
is not the decision-maker. The PC is the decision maker, and
the spreadsheet a tool that  can inform same. So, if there is
a proposal that we bung this out and just haggle all night on
the phone, that's completely in order.

Let's look at the spreadsheet and read through the first row,
which happens to be the e4-related talks. We will skip to
column D.

D contains the amount of submissions that have that tag
associated with them, *after* we have gone through and
'cleaned' up the tags. There is overlap. Here we see that
there are a total of 24 submissions that are tagged e4.

E contains the % of the overall cleaned submissions that
is represented by the number in D. You can see the e4
tag represents 3.99% of submissions.

For guidance - yellow columns indicate percentages of
conference time. I will come to the other two later.

The blue columns F-I contain allocation numbers, which
indicate the number of slots that are available for that
tag. Here's how to read it for e4:

F -- e4 gets one talk minimum, as does everything else.
Change this column to up the value for more topical or
interesting tags - the figures will adjust.

G -- after the mandatory talk slots from F are subtracted
from the total talk slots (see E1) the remaining slots are
divided up pro-rata over the tags using the figure in E -
in this case, the 3.99% submission rate achieved by e4
means that tag earns 1.95 slots.

Yes, these figures are decimals for convenience, you can't
have 0.05 of a talk, but it just allows us to see which way
to go when rounding.

If there are any special adjustments to the time, these
are added in H. If you look down the H column until
rows 22 and 23, you will see an example of an adjustment
in use - there are zero web services and data tools standard
talk requests.

Finally, the I column sums the previous three to show the
number of slots allocated. For e4, this comes into 2.95
standard talks.

J is another yellow column, which means it is a percentage
of conference time. Here the 2.95 from I is converted into
a percentage of the overall time available for standard
talks. This number tells us that with that number of slots,
e4 is starting with 4.34% of the time available to all standard
talks.

A brief interlude to explain the number of slots in E1. We
started out with 80 and now we have 68. What has happened?
I've completely excluded the OSGi DevCon allocation here,
since that is a known and set allocation. I've also removed
the Director's Choice slots too, the maths are 80-8-4 = 68.

The purplish columns are the ideal world, where we have
space for every single standard talk application. Column L
is the amount of standard talks actually requested for that
particular tag. This number comes from the 'haggling-standard...'
document (however it is not live-linked). The K column
contains how many slots would be allocated in the ideal
world, where we could fit everything, based upon the
submission levels. It's really just for illustration purposes,
so I hope it isn't distracting.

The M column shows just how much of a reduction would
be needed to line everything up, if a spreadsheet was deciding,
which it isn't.

The N column is once again a percentage of conference
time column. In this case it shows the percentage of slots
that has been requested by a particular tag. In the case
of e4, its 10.68%. I arrived at this figure based upon the
content of column L, which comes from the haggling document.

Finally, we have the usual subscription numbers. Column O
tells us that e4 has asked for 245.88% more time than it has
been allocated to start with. Column P tells me that e4 has
asked for 267.88% more time than it would get if it was based
totally on the number of talks submitted with e4 on them.

The columns that you need to pay most attention to are

Column I :  using the basic algorithm, this is how many talk
slots will be allocated.

Column L:  this is the number of talks you were looking for.

Column M: this is the number of talks IN THEORY you would
need to cut if the spreadsheet was doing the thinking.

Column N:  the is the relation between the number of slots
allocated and the number of slots requested. These tags
need more slots!

Well, that was a lot of dumpage. I hope most of it came
through, especially the bit about the PC making the decisions.

 cheers
   --oh
_______________________________________________
eclipse.org-eclipsecon-program-committee mailing list
eclipse.org-eclipsecon-program-committee@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/eclipse.org-eclipsecon-program-comm
ittee