Suggest some improvements!
- Thread starter Morrus
- Start date
InShambles
First Post
Fix the messageboard clock.
Review Scores
I would suggest basing the overall review score for products on the... oooh, I always get these wrong... median (I think) rather than the mean. The one where you drop the highest and the lowest and average the ones in-between. That way, one crap review couldn't tank a great product and one super pie-in-the-sky review couldn't pull a lousy product out of the "stay away at all costs" territory. There's always one person out there willing to write a completely off-base review just to affect the score. Two are harder to find.
And, for the record, this is something I've always thought would be a good idea, before I had a book reviewed here.
AJL
I would suggest basing the overall review score for products on the... oooh, I always get these wrong... median (I think) rather than the mean. The one where you drop the highest and the lowest and average the ones in-between. That way, one crap review couldn't tank a great product and one super pie-in-the-sky review couldn't pull a lousy product out of the "stay away at all costs" territory. There's always one person out there willing to write a completely off-base review just to affect the score. Two are harder to find.
And, for the record, this is something I've always thought would be a good idea, before I had a book reviewed here.
AJL
Blacksway
Rock Monkey
Review system suggestions:
Ranking (tleilaxu's suggestion):
This relies on a) the rating system meaning more than it does, and by this I mean that at the moment we use a 1-5 system, which gives some indication of how much someone liked or disliked a product, but it is the actual review you should be worried about not the rating. Do have something being better than X% of other supplements doesn't really mean that much and further pushes the rating to the fore rather than the review. This is only my opinion on the whole rating/percentage/out of 10/5 thing of course. It also would reply on b) johnsemlak's suggestion...
Categories (johnsemlak's suggestion):
This one is a good suggestion in priniciple, unfortunately it suffers from many practical problems. 1) it is technically complex, 2) who decides what the categories are/called? 3) who goes through putting all the products into the categories? Trust me - if we run out of other things to do then we'll serisouly consider this, just don't hold your breath
Search enhancements (the other half of johnsemlak's suggestion):
There are some search enhancements in the pipeline
Average not being average? (AaronLoeb's suggestion):
Firstly, what your talking about is not median either (median is where you put all the number in a row, in order and pick the one in the middle, e.g. the median of 1 1 1 4 4 is 1 (the third number of five), the mean is 2.2 (11/5))
Secondly, I don't think it would work - a) because as I mention above the curent rating is simply an estimate of how much the user liked/diskliked the product, hence the average is just a general indication of average opinion, excluding some because they happen to disagree with someone else is surely not fair? b) how do you decide which ones to exclude? say you only have three reviews, 1, 4 and 5. Should your new average exclude both 1 and 5? giving an average of 4, but based on only 1 review? is an average of 3.3 really that different, and perhaps more accurate because its based on 3 reviews? Who says the people who scored it at 4 and 5 are right? and the one on 1 wrong? its only their opinion after all. If you really want to find out what the product is like, then read the reviews and find out WHY they liked/disliked the product.
Well enough of my ranting. If you have any comments on a particular one of these issues then start another thread (or email me), and leave this one for more suggested improvements, for both messageboards, news and reviews!
Oh, and my new email address is sam@blacksway.net for those who have my old blueyonder one.
Ranking (tleilaxu's suggestion):
This relies on a) the rating system meaning more than it does, and by this I mean that at the moment we use a 1-5 system, which gives some indication of how much someone liked or disliked a product, but it is the actual review you should be worried about not the rating. Do have something being better than X% of other supplements doesn't really mean that much and further pushes the rating to the fore rather than the review. This is only my opinion on the whole rating/percentage/out of 10/5 thing of course. It also would reply on b) johnsemlak's suggestion...
Categories (johnsemlak's suggestion):
This one is a good suggestion in priniciple, unfortunately it suffers from many practical problems. 1) it is technically complex, 2) who decides what the categories are/called? 3) who goes through putting all the products into the categories? Trust me - if we run out of other things to do then we'll serisouly consider this, just don't hold your breath
Search enhancements (the other half of johnsemlak's suggestion):
There are some search enhancements in the pipeline
Average not being average? (AaronLoeb's suggestion):
Firstly, what your talking about is not median either (median is where you put all the number in a row, in order and pick the one in the middle, e.g. the median of 1 1 1 4 4 is 1 (the third number of five), the mean is 2.2 (11/5))
Secondly, I don't think it would work - a) because as I mention above the curent rating is simply an estimate of how much the user liked/diskliked the product, hence the average is just a general indication of average opinion, excluding some because they happen to disagree with someone else is surely not fair? b) how do you decide which ones to exclude? say you only have three reviews, 1, 4 and 5. Should your new average exclude both 1 and 5? giving an average of 4, but based on only 1 review? is an average of 3.3 really that different, and perhaps more accurate because its based on 3 reviews? Who says the people who scored it at 4 and 5 are right? and the one on 1 wrong? its only their opinion after all. If you really want to find out what the product is like, then read the reviews and find out WHY they liked/disliked the product.
Well enough of my ranting. If you have any comments on a particular one of these issues then start another thread (or email me), and leave this one for more suggested improvements, for both messageboards, news and reviews!
Oh, and my new email address is sam@blacksway.net for those who have my old blueyonder one.
Last edited:
Mark
CreativeMountainGames.com
Is there anything that can be done (and I mean by this, practically done) that would allow for search words to be as few as three letters? Searching by even the term D&D or d20 will halt the search with the warning that you need four letter words or higher. I wanted to do a search by the word "war" the other day, which seemed reasonable. Under the current situation you cannot search by thread titles for company acronyms either, which is somewhat problematic, I think. Just a thought for the in box pile. Thanks to Blacksway, Morrus, Piratecat and all of the mods, admins and members who help everyday to make these boards continuously useful and helpful to all of the rest of us knuckleheads...
Liquide
DEX: 4
Mark said:Is there anything that can be done (and I mean by this, practically done) that would allow for search words to be as few as three letters? Searching by even the term D&D or d20 will halt the search with the warning that you need four letter words or higher. I wanted to do a search by the word "war" the other day, which seemed reasonable. Under the current situation you cannot search by thread titles for company acronyms either, which is somewhat problematic, I think. Just a thought for the in box pile. Thanks to Blacksway, Morrus, Piratecat and all of the mods, admins and members who help everyday to make these boards continuously useful and helpful to all of the rest of us knuckleheads...
Even I can answer this one, the reason why you cannot searh for a string shorter then 3 letters is due to clogging and the wildcards.
When searching for three letter it looks for that sequence in all words and all instances in the database you specifiy (in this case description, publisher and product title fields).
And in order for it to work pleaseble fast (and not clog the server badly) it was capped (orignially by me) at 4 letters minimum.
If BlackSway however can input some specific 3 letter combos that are allowed in a seperate database table this can be done, but is no a direction you want to take with a search function.
Just my 2 cents.
Liquide
DEX: 4
Mark said:
much lesser chance of a 4 letter series to come up with results that aren't wanted then 3 letter series.
3 letters tend to mess up a bit (and I would like to see someone search for "and" for example).
It is a level of security, not to clog up and make the server come to a standstill. Also the common way of coding a search that CAN have hundreds of multiple searches (or more) at the same time.
As long as you keep them simple they do not cause you headaches, when you start to make them more advanced/allow for shorter words/letter combos the server starts to rapidly clog up and come to a standstill.
