2011-12-30, 08:21
Ned, I had a couple of thoughts on the wiki that I wanted to share. I am sure some or all of these things have been thought of, but for the sake of clarity and progress I wanted to spell them out. I've worked in technical support roles for a long time now, and one of the first things you learn is to cater to the lowest common denominator. Some people are (very) offended by this mentality, but it has been proven time and again that if you assume your user understands something that you have not spelled out for them, you are going to spend twice as much time dealing with their problems. Even learning this from experience, I often catch myself making assumptions that people understand the fundamentals behind a subject we are discussing in a forum (not just here), and it backfires on me all the time.
One of the biggest problems with any wiki, as mentioned earlier in the thread, is a lot of people do not like to use them or simply WON'T use them. I think there are a number of factors for this attitude, but one of the biggest is that a wiki can store a huge amount of information, but does not typically provide a simple and intuitive way to retrieve and view the data. Category pages are overwhelming, especially to new users, if they have dozens and dozens of links on them. To make it more complicated, the links are just vanilla titles with no real description which can be very unhelpful in certain circumstances where you may not know exactly what you need to look at (new users, again).
Another example of how unhelpful a wiki can be is when you use the search function. When you search for something you expect to get at least a couple of results. With the xbmc wiki, you tend to get nothing at all with more generic queries--at first. If you are very specific and your query has a a dedicated page of the same name (i.e. advancedsettings.xml) you are taken directly to the page. However, search for something more vague such as "MySQL errors" and you get nada:
On that page if you click the Check "All" button and search again, you will get several results. That is good, however a lot of people won't intuit that and will be frustrated that a "basic" query seems to lead to zero results most of the time.
Admittedly, using the check all option will produce results that may be rather generic, however it demonstrates that the search feature DOES work. When you keep getting blank results, it only takes 2 or 3 of those before you start wondering if the search function actually works at all. If you are new to wiki's in particular, or not very internet/technically savy, you are going to give up on that first page and run to the forum instead. Users, in general, do not want to spend hours reading through a bunch of stuff that explains every little detail that may or may not be directly related to their problem. They want answers and they want them now. That means they have to manage a search that produces results that QUICKLY leads them to their solution, OR they are going to post on the forum and wait for someone to spell it out for them. The people that like getting under the hood and digging through things really are a minority group.
Personally, I use Google's site search function for just about any searches I make on any website. I am very comfortable with Google search, I know what to expect from Google, and it's dang hard to beat a search engine that has spent billions of dollars on updating and maintaining it's search features. For example, using the same MySQL errors query with Google site search:
Google wiki search.
In this case, I am using a very generic query which also does not have a lot of information spread throughout the wiki, so I only get 14 results (google hides 3). The wiki search only produced 8 results with the same query. I prefer to have as many results as possible returned, regardless of how vague they may be, if I am searching for something that is poorly documented-- but that's just me.
Regardless of the search engine you use, coming up with zero results on a regular basis WILL deter users from actually using the wiki, so anything you can do to prevent that would probably be beneficial to your end goal here. Over the years, especially on forums, I have come to realize that many (perhaps most?) people are simply not good at crafting GOOD search queries, which naturally leads to people asking the same questions over and over again. Some people don't even bother with a search, but I don't have a good solution for that except a 3' long 2x4 to the back of the head.
I'm not suggesting that catering to lazy users is necessarily the "right" or "best" way to approach things, but I AM suggesting that if you want the wiki to be more widely adopted and you want to stop seeing the same questions in the forums as often, you WILL have to make things as easy as humanly possible because people are just fricken lazy--and often stupid (XBMC doesn't have any stupid users though. I'm just saying "users" in general...)
If the thing you are interested in learning about is not directly mentioned and linked on the main page, and you try using the search and get no results, how does an "average user" find what they are looking for? I believe the answer to that is they either give up, come to the forum instead, or try their luck with a WWW search on Google (or whatever search engine they favor). In that case, if there WAS a page with a lot of good info on the subject they were looking for, they will have missed it (unless someone points them to the page) and the wiki has not done it's job.
My next comment relates to outdated information. The very nature of a wiki makes it impossible to keep everything updated, and the deprecated information from being confused with the current versions. For instance, if someone had been using material that referenced the keymaps.xml discussed earlier in the thread on their pre-eden build, they would run into problems and not understand why (again leading to a possible repeat question in the forum). In a case such as this, it would be ideal to have a very obvious button or link for anyone to click on to flag the page as outdated, which would then alert someone on the wiki staff so they could verify this and either alert users that the information is no longer relevant/accurate and is flagged for update when someone gets to it or move it to an "archive" and replace it with a generic placeholder and request for current version specific information from the community--along with a link to the old information so people at least have a place to START until new information is added for the current version.
Obviously, it's probably best to keep the old data around for people that are still running old versions for whatever reason. However, the mix of old and new information is VERY confusing for many users, and could potentially lead to major mistakes being made with their installs. It would be ideal if these pages could not only be updated for current builds, but also "archive" the older information to help avoid confusion AND provide some form of legacy support. The "archived" pages should be identified by xbmc specific builds, and should be very obviously different from current build information pages. Perhaps a color coded banner of some sort at the top of each page with the build name included. Then you could have a link to the "archived" wiki on the main page (for easy legacy support) which would eventually be a mirror of the current wiki but only containing information for legacy versions as well as provide the legacy users a clear and easy way to find what they need without adding as much confusion for the current version users.
Lastly, wiki pages that contain instructions for multiple operating systems on the same page CAN be confusing if they are not very carefully formatted. Often times these types of pages will have the first few steps outlined as universal steps. Then in the middle, things will start branching off for different OS'es. At this point, it can quickly get overly complicated to follow these instructions, especially if the instructions branch by OS, then merge back to a universal format, then branch again, and so on. The simple solution for this is to try and implement a template for instructions that uses something like a simple color coded border around code boxes for each OS. If each OS consistently uses the same color coded boxes for OS specific steps, then users can quickly identify the steps that apply to them and ignore the rest without confusion. No color code means it is a step that you need to follow, regardless of OS.
Of course, the biggest hurdles for the last 2 issues I raised are implementing the features with wiki editors adhering to the systems, and of course the huge task of modifying the content already in place. I'm not saying all of this is 100% perfect or that I have all the answers, these are just observations that I thought might be productive to the overall user satisfaction of the wiki. Implementing an archive feature alone would be a massive project, so I realize it is highly unlikely. Regardless, all of the work that has been done on the wiki is a great service to this community and I am sure everyone appreciates the effort.
Apologies for writing such a long diatribe. I have tried consolidating my thoughts some, but I guess I am too tired to articulate my thoughts with fewer words tonight
One of the biggest problems with any wiki, as mentioned earlier in the thread, is a lot of people do not like to use them or simply WON'T use them. I think there are a number of factors for this attitude, but one of the biggest is that a wiki can store a huge amount of information, but does not typically provide a simple and intuitive way to retrieve and view the data. Category pages are overwhelming, especially to new users, if they have dozens and dozens of links on them. To make it more complicated, the links are just vanilla titles with no real description which can be very unhelpful in certain circumstances where you may not know exactly what you need to look at (new users, again).
Another example of how unhelpful a wiki can be is when you use the search function. When you search for something you expect to get at least a couple of results. With the xbmc wiki, you tend to get nothing at all with more generic queries--at first. If you are very specific and your query has a a dedicated page of the same name (i.e. advancedsettings.xml) you are taken directly to the page. However, search for something more vague such as "MySQL errors" and you get nada:
On that page if you click the Check "All" button and search again, you will get several results. That is good, however a lot of people won't intuit that and will be frustrated that a "basic" query seems to lead to zero results most of the time.
Admittedly, using the check all option will produce results that may be rather generic, however it demonstrates that the search feature DOES work. When you keep getting blank results, it only takes 2 or 3 of those before you start wondering if the search function actually works at all. If you are new to wiki's in particular, or not very internet/technically savy, you are going to give up on that first page and run to the forum instead. Users, in general, do not want to spend hours reading through a bunch of stuff that explains every little detail that may or may not be directly related to their problem. They want answers and they want them now. That means they have to manage a search that produces results that QUICKLY leads them to their solution, OR they are going to post on the forum and wait for someone to spell it out for them. The people that like getting under the hood and digging through things really are a minority group.
Personally, I use Google's site search function for just about any searches I make on any website. I am very comfortable with Google search, I know what to expect from Google, and it's dang hard to beat a search engine that has spent billions of dollars on updating and maintaining it's search features. For example, using the same MySQL errors query with Google site search:
Google wiki search.
In this case, I am using a very generic query which also does not have a lot of information spread throughout the wiki, so I only get 14 results (google hides 3). The wiki search only produced 8 results with the same query. I prefer to have as many results as possible returned, regardless of how vague they may be, if I am searching for something that is poorly documented-- but that's just me.
Regardless of the search engine you use, coming up with zero results on a regular basis WILL deter users from actually using the wiki, so anything you can do to prevent that would probably be beneficial to your end goal here. Over the years, especially on forums, I have come to realize that many (perhaps most?) people are simply not good at crafting GOOD search queries, which naturally leads to people asking the same questions over and over again. Some people don't even bother with a search, but I don't have a good solution for that except a 3' long 2x4 to the back of the head.
I'm not suggesting that catering to lazy users is necessarily the "right" or "best" way to approach things, but I AM suggesting that if you want the wiki to be more widely adopted and you want to stop seeing the same questions in the forums as often, you WILL have to make things as easy as humanly possible because people are just fricken lazy--and often stupid (XBMC doesn't have any stupid users though. I'm just saying "users" in general...)
If the thing you are interested in learning about is not directly mentioned and linked on the main page, and you try using the search and get no results, how does an "average user" find what they are looking for? I believe the answer to that is they either give up, come to the forum instead, or try their luck with a WWW search on Google (or whatever search engine they favor). In that case, if there WAS a page with a lot of good info on the subject they were looking for, they will have missed it (unless someone points them to the page) and the wiki has not done it's job.
My next comment relates to outdated information. The very nature of a wiki makes it impossible to keep everything updated, and the deprecated information from being confused with the current versions. For instance, if someone had been using material that referenced the keymaps.xml discussed earlier in the thread on their pre-eden build, they would run into problems and not understand why (again leading to a possible repeat question in the forum). In a case such as this, it would be ideal to have a very obvious button or link for anyone to click on to flag the page as outdated, which would then alert someone on the wiki staff so they could verify this and either alert users that the information is no longer relevant/accurate and is flagged for update when someone gets to it or move it to an "archive" and replace it with a generic placeholder and request for current version specific information from the community--along with a link to the old information so people at least have a place to START until new information is added for the current version.
Obviously, it's probably best to keep the old data around for people that are still running old versions for whatever reason. However, the mix of old and new information is VERY confusing for many users, and could potentially lead to major mistakes being made with their installs. It would be ideal if these pages could not only be updated for current builds, but also "archive" the older information to help avoid confusion AND provide some form of legacy support. The "archived" pages should be identified by xbmc specific builds, and should be very obviously different from current build information pages. Perhaps a color coded banner of some sort at the top of each page with the build name included. Then you could have a link to the "archived" wiki on the main page (for easy legacy support) which would eventually be a mirror of the current wiki but only containing information for legacy versions as well as provide the legacy users a clear and easy way to find what they need without adding as much confusion for the current version users.
Lastly, wiki pages that contain instructions for multiple operating systems on the same page CAN be confusing if they are not very carefully formatted. Often times these types of pages will have the first few steps outlined as universal steps. Then in the middle, things will start branching off for different OS'es. At this point, it can quickly get overly complicated to follow these instructions, especially if the instructions branch by OS, then merge back to a universal format, then branch again, and so on. The simple solution for this is to try and implement a template for instructions that uses something like a simple color coded border around code boxes for each OS. If each OS consistently uses the same color coded boxes for OS specific steps, then users can quickly identify the steps that apply to them and ignore the rest without confusion. No color code means it is a step that you need to follow, regardless of OS.
Of course, the biggest hurdles for the last 2 issues I raised are implementing the features with wiki editors adhering to the systems, and of course the huge task of modifying the content already in place. I'm not saying all of this is 100% perfect or that I have all the answers, these are just observations that I thought might be productive to the overall user satisfaction of the wiki. Implementing an archive feature alone would be a massive project, so I realize it is highly unlikely. Regardless, all of the work that has been done on the wiki is a great service to this community and I am sure everyone appreciates the effort.
Apologies for writing such a long diatribe. I have tried consolidating my thoughts some, but I guess I am too tired to articulate my thoughts with fewer words tonight
