If you register for free, you will be able to post threads, vote on polls and lots more. If you have problems with the registration or logging in, please contact the administrator.
AI Experiments – A Guide to Starting Up
<UL TYPE=SQUARE>
<LI>Make the aim of the experiment clear and original – Indicate in your opening post exactly what aspect(s) of the AI you are looking to examine. Are you investigating a particular angle such as squad management, finances, or personality? Are you doing it based upon default or constructed data? Make sure your idea is original, and proof read your openers before you post them – they are critical.
<LI>Ensure that you are interested in the experiment – Don’t create an AI experiment for the sake of it – make sure that you have an interest in the outcome. Writing reports is a time consuming commitment, and if you lose interest the experiment will quickly fade. Also, if you like it, chances are someone else will.
<LI>If you’ve edited data, explain exactly how – If you don’t do this, you’ll answer the same questions over and over. Give people all the information they need to understand the dataset of the experiment – as much as anything they then have the knowledge to contribute to explaining whatever you see. If you’ve changed a number anywhere in the database, say so.
<LI>Clarify the manager status – Are you playing in the game, and therefore making it less of an AI experiment? Are you examining how the AI responds to certain actions you take? Or is this a holiday game, where the AI is in charge of everything?
<LI>Think about the structure of your game – It is advisable, in a holiday game, to set the detail level of every match that comes under the experiment umbrella, to full. Think about which leagues you want to run – this is a personal decision based upon what you are trying to achieve. If you examining the career of a player, you may wish to play more leagues to give him more chance to roam the world. If you are following the progress of a constructed team, perhaps only their country is required. But again, state and explain your decision.
<LI>Update regularly – Give regular and structured updates. In a long term holiday game, these updates may be seasonal. In a shorter term experiment, updates may be more regular. Try to strike a balance between updating regularly enough to provide all relevant information, but far enough apart to let the experiment progress. The regular update is the most important part of the experiment – be prepared to put the time in to give all the information your readers request, if relevant. This helps provide an immersive experience.
<LI>Use Genie Scout – Much of the AI in the game can be explained using figures that you wouldn’t otherwise see. Club and player decisions are often based on hidden attributes such as mental traits and reputation.
<LI>Include plenty of screenshots – A picture says a thousand words. Star players and league tables are especially important here – you cannot list the stats of every player that comes along so provide screenshots. If people aren’t interested, they won’t click, and it helps keep the report concise and readable while still providing maximum information.
<LI>Be prepared to modify the experiment aims as you go – You may discover an AI oddity that requires further exploration. The game world may also progress to a point where some or all of your original aims may be obsolete. Be prepared to examine new aspects of the AI as you discover them, it helps the experiment progress.
<LI>Be prepared to modify the structure of your regular report – You may find yourself receiving requests about the game world, especially if your experiment continues for decades. Consider introducing new sections to your regular reports to cover these – you never know what information others find interesting. Remember your reports are for your readers, otherwise you’d be doing the experiment alone.
<LI>Be friendly – You may have already answered the question on page six, but not all your readers will read every post. Just answer it again. Putting as much information as possible in your opening post helps too. Try to answer all questions – your readers have as much to contribute to the experiment as you, and if you don’t answer their queries about the game, they’ll lose interest. Remember it is easier for you to be immersed in the game – you have the save.
</UL>
AI Experiment Guide/FAQ 
Linear Mode
