Every day is like Sunday...

I'm having some issues here. (warning: major geek fest approaching)
The dataset I'm supposed to be doing, that the project will be based on, and which I thought was complete some weeks ago, is undergoing some "reimagining". Turns out the government goons that are funding the whole thing think that there needs to be more Pacific conflict in the dataset, and therefore I must include Fiji, Vanuatu, New Caledonia and the Solomon Islands. Fine, they've experienced conflict in various forms, and nobody has collected any information on them, so its up to me.
So thats all well and good.
But.
In order to include these countries, I have to change a few rules, so that the dataset has definable criteria for inclusion. Like, lowering the threshold of battle deaths from 25 to 10. Fine, theres Fiji taken care of. Removing the rule that Side A of each conflict must be the government. Cool, there goes the Solomons in the dataset. Change the definition from civil war to internal violent conflict, that way we can include the hostage incident in New Caledonia, and now they're in. Yay.
So now that we've got new rules for the set, I'm going to have to go back through the original set and code every incident in the other nine countries that qualifies as internal violent conflict, which now includes incidences of riots, coups and insurgencies resulting in ten battle deaths or more. Fine.
I also have to recode the conflicts themselves, because the identification codes we took from the PRIO set don't include IDs for the Pacific countries, because they use the 25 battle deaths threshold, and the involvement of the government as criteria. I'm stealing, sorry, borrowing, country codes from Correlates of War, and adding a specific code for each conflict that I'm making up.
Plus a whole bunch of other codes, like subID, type and incompatibility will need fixing, because each new conflict, being that by inclusion is no longer necessarily a civil war, will need to be coded as war, violent conflict, insurgency, coup, riots etc, and will also need to be defined as ethnic, territorial, political or other. Super fantastic.
However, no matter how much I try and bend the rules, I'm absolutely buggered if I can find a way to include Vanuatu in the dataset, given that in each incidence of conflict as far as I can find out, only one person died. If it was, say, 9, then I could just assume that whoever counts these things missed one person, and then I could include them. But the difference between nine dead people and one dead person is pretty significant. (not that I wished more people had died!)
There is also a problem with Fiji. (in that I'm stuck in rainy Christchurch and am not there lounging on a beach with Joey Rokocoko rubbing my feet and feeding me mangoes).
The 1987 coup is highly significant regionally, in that it demonstrated deepseated ethnic tensions, a hangover from colonial times. Nobody died though to the best of my knowledge, which yay is good, but it means that it can't be included in the project, implying that Fiji has no problems, aside from the 2000 coup and subsequent mutiny. Which it clearly does. (One of which being that I'm not there!!)
Thats the problem with this social science business. You can't just include something cos you think it should be there, not like the good old subjective humanities. No matter how much you want to, cos then other people who know more about this than me are going to do all sorts of flash statistical shit on it, and if it all goes pearshaped, then I'm in the poo. Plus there is the chance that the people who make the policy on this stuff are going to use the results of all our research, and I can't have them blaming the next Fiji coup or the failure of the New Caledonian independence movement on me, a lowly researcher.
I know I can fix all these problems, and if I got my shit together I could even have the complete dataset by Christmas. The thing is, I have a meeting at 10am tomorrow to clear this up, and the powers that be want a complete set by then. Its currently 5.20pm, I start work at the restaurant at 6pm, and I'm still in my dressing gown, having spent the whole bloody day trying to finish the dataset.
I was supposed to spend this weekend sorting out my life! Oh well, its waited 28 years to be sorted out, I'm sure one more week won't hurt.
Normal programming will resume....um....eventually.
Poo is a funny word.