|
fcharton -> RE: Perfection, of a kind, spence (A) vs fcharton (J) (11/3/2013 9:56:42 AM)
|
January 26th 1943
Sorry about the long interruption. Work was a bit crazy over the last two weeks. A large audience survey (this is what my day job is about) was due to be published next week. The person who usually checks and processes the data (and compensates for the lacks of the pollsters and survey designers) had been on sick leave for several months. And unfortunatelly, he does all his processing in APL, a marvelous language, really, but quite hard to read if you’re not very good in maths and data processing (it was once derided as a “write only language”).
So, after months of procrastinating and mellow excuses (“oh we’re just a few days late”, “right, I think I understand the process now”, “everything is under control”,” it is going to be all right”, “I’m confident”), the wheels began to come off three weeks ago, when it was noticed that one statistical routine (something that performs a discriminant analysis, as part of a data fusion process) didn’t work this year, for reasons no one understood. And so I wrote a new routine to process the data. So far, it was fun…
Then, after a few nights spent trying to figure out why an apparently very simple process didn’t work as advertised, it became clear that we didn’t understand the APL programs we were supposed to use, and that the myriad details no one had deemed necessary to check hadn’t taken care of themselves. And of course, this meant every thing had to be programmed from scratch. And checked too, because it now was clear most of the people involved, pollsters and designers, had thought someone would do their dirty jobs for them (young university graduates, fat resumes over empty heads…)
And so we did it the old way, with spreadsheets checked line after line for errors, and ugly C programs that do the job, and do it quick. And it took two weeks, days, nights and weekends. And yesterday evening, the files were ready at last, and I opened a bottle of Rioja, and sent back my first AE turn in almost two weeks.
Now, I do like data processing and I think it is something I do well, but this will probably be my last survey processing. It is very hard work, but way too underappreciated, and you often end up the scapegoat for every one else in the project. And the funny thing is, the better you do your job, the more tempting it becomes for others not to do theirs. This is what happened to this survey. Since my colleague was there to check and correct everyone’s errors, the data got worse and worse over the years.
Why am I telling you all that? First, because it makes me feel good (hey, that’s MY AAR!), and anyway I have little to say about the turn today, which I played fast, just to give my opponent a sign of life (btw, Spence really is a nice opponent, I’ve seen he advertised for an opponent, go for him if you want a good game…), but also because I thought a lot of AE, and my misgivings about the game, during those two crazy weeks.
First, the way my survey data got wrong has a lot in common with the way AE games can turn bad. There are lots of unimportant details in the game that can easily be left aside, and will eventually accumulate. I have not been taking care of my empire for months, and it is beginning to show. This happens on several levels. There are, of course, basic tasks that get delayed, units staying in the rear, tankers not moving, bases not being built. But more importantly, I lack a method on many aspects of the game. I have not thought yet about pilot training, or search areas, or AA, or kill zones for my LBA. There are just too many aspects in the game.
This brings another similarity with data processing. You easily get lost in details, and fail to see the general picture. In AE terms, this might be described as focusing on the tactics instead of the strategy, or on the war instead of the logistics. In data processing, you usually know you’re on the right track when you are able to write down the process as a list of a few meaningful tasks (which you can then refine into subtasks, and so on). This is what you learn in humanities, when they teach you to write essays, and especially to write plans (the old fashioned ones, with three parts, not the bullet list stuff we get nowadays). This is something I need to do in AE.
Finally, one point AE and data processing have in common is that you pretty much learn everything through practice. I know some players like to believe their historical reading, or their military experience, or both, give them an edge, or turn them into “naturals”. I don’t think so. You need a special set of mind to get interesting in AE, obviously, but apart from that, you become a good player because you play a lot, and because you pay attention. Data processing (most computer programming in fact) works exactly the same, in the word of one of my former directors: “when you have made as many errors as me, you will be the boss”.
So, I’m back into the game, this short work craze made me realize I wanted to go on with it, and I realize I need a plan. I’m in the end of January 43, and have reached my 1942 objectives month ago, but I’m clueless about what to do. The Allies are not on the offensive yet, and I can probably make a difference now. Let’s be about it, then.
Near Chittagong, it turned out that an air support unit actually pursued my retreating troops. We attacked it today, and destroyed it.
Ground combat at 56,42 (near Chittagong)
Japanese Deliberate attack
Attacking force 63900 troops, 981 guns, 1527 vehicles, Assault Value = 2232
Defending force 850 troops, 0 guns, 0 vehicles, Assault Value = 1
Japanese assault odds: 1456 to 1
Allied ground losses:
1054 casualties reported
Squads: 0 destroyed, 0 disabled
Non Combat: 113 destroyed, 0 disabled
Engineers: 0 destroyed, 0 disabled
Units destroyed 1
Defending units:
172 Wing
I believe this is a bug in the code. Pursuit, involving movement, should be limited to combat units, and most probably motorized ones. It is not quite the detail, as pursuit entails heavier casualties for the defeated side.
In the air, over Dobo, we traded seven Oscars for three Kittihawk III and a Spitfire Vc. Over Rabaul, it was four Nicks for a B-17E, at night. Night battles against 4E are extremely bloody for the Japanese, and I probably need to commit cheaper planes (A6M2) to it.
Singapore now has 40k tons of supplies. We’re getting there.
And Dobo held another day.
This is all for today. I was tired yesterday, and not too confident on my ability to make the correct decisions after the Rioja (Alcorta 2008, a pretty good wine, I really like Spanish wines). But I’ll be back tomorrow, with a plan, or at least the premises of one.
|
|
|
|
