The guide explaining the details of the database, distributed to the head researchers by SI. I think either Graeme Kelly or Phil Rolfe did the first version of it, borrowing heavily from an old CM researchers' guide for the format and some of the text; I revised it a couple of times after that in subsequent years, clarifying and correcting some things and also pointing out db-related coding changes that the HR's needed to be aware of.selne wrote:1. You are very unprecise, JeffR. What's the researcher's guide?
And you've done quite a good job putting it all together; I suspect the problems I see stem from people with a copy of the guide posting partial or superseded information from it when they were trying to explain ratings to those who hadn't worked on the original research. The "Estimated attribute values by CA" table, for example, was something I included as a quick reference for the researchers to illustrate a rule-of-thumb I mentioned at the end of a two-page discussion of the CA changes. Unfortunately, whoever first copied it out of the guide and posted it just gave the table and not the real meat of the text that explained why the numbers now worked that way, and how it was different from the code in previous editions. That's what I was trying to explain above.selne wrote:2. It's been a lot of work to make that guide, i said in the beginning that there were helpers involved (which is simply taking over a few explanations of attributes from the tbn site). But still most of the guide is original content with just very few quotes. Bruins72 asked me if i could do some sort of guide, and he helped me a lot with suggestions to improve its content and the laguage errors.
Sure - if it had been noticed. A great many things were caught and fixed in testing, but that wasn't one of them. I'm embarrassed it slipped by me in particular because that's exactly the sort of thing I looked for in my testing, but I never caught it.selne wrote:Why don't you name exactly what's so off the mark. Then we talk about it. We can improve the guide anytime, if you have sugggestions, please share them with us. If you know the code, then i have to tell you, you better take a closer look at it. Because the programmers made the influence of weight too big, though not on porpuse. But they could have corrected it while testing.
It's quite wrong for EHM07, though, and, as I said, will create problems in the database if it's done on a large scale. I don't think anyone wants to see that happen to the work Lidas or anyone else is doing, and that's why I wanted to point it out when I saw people taking the direct CA-to-attribute approximations seriously earlier in this thread.selne wrote:The quote on CA avarages from Alessandro was put into the guide because it's very easy to understand for newbies about how the game calculates the attys. It's just for the first impression.
When I was doing the last revision of the research guide, I spent quite a while talking back-and-forth with Riz about what he'd changed with CA to make sure I understood it properly and could explain it to the other HR's, in the hope that everybody would get the message. That was an unusual amount of time for him to spend on db-related stuff, but it was that important. Unfortunately, while some of the guys paid attention to the big red "read me, it's important" text in their new guide - Matt Bosela's one I remember getting it exactly right - not all did, and there wasn't time to fix all of the problems. But no reason to keep repeating their mistakes when it can be clarified now, right? Much easier to avoid people creating bad data in the first place than it is to find and fix the problems later - believe me, I know, after spending hundreds of hours trying to undo damage done to some of the league db's, with only mixed success.
Re: sharing the guides, as much as I'd like to, their content, even the stuff written by me, ultimately belongs to SI and just isn't mine to share. I'll happily clarify db-related stuff where I can, but posting old EHM work product wholesale would be more than a little out-of-bounds.
