It sounds like your problem is a special case of a more general problem: Your client wants you to implement a system that you don't expect will work well, as specified, based on your superior knowledge of the system's domain and technology.

This question on The Workplace Stack Exchange about a graphic designer dealing tactfully with a boss who is ignorant of graphic design principles seems pretty parallel, and the answers there offer some advice that can be adapted to your situation.

• Ask lots of leading questions.

  Although your client has autocratic control of the project, its requirements, and her ultimate behavior as a moderator, you have some leverage, as the person implementing the project, to tweak the technical solution and also nudge your client's thinking toward a system behavior that will work better. You can ask questions about how she wants certain aspects of the project implemented in ways that bring up issues you are concerned about. For example:

  Should I add a filter that automatically rejects submissions that don't include a TPS number? What should it tell the user about why that number is necessary?

  How do you envision the review/improvement cycle working? Should I include a messaging system that you would use to explain what work is needed to improve a submission before you'll accept it?

What do you think of this streamlined version of the entry form?

• Introduce third-party sources.

  Your client is apparently uninterested in learning about what it's like to moderate an online forum, so it may be difficult to get her to listen to third-party sources. However, you may be able to sneak them through the door by bringing them in to help with your implementation. For example, say that you and the rest of the implementation team want to get some fresh ideas about designing a forum like this, and bring in someone with experience doing so to give a little seminar to the team, and invite your client to attend. Similarly, you could read a book or authoritative online resource, ostensibly for the purpose of getting implementation advice, and then "discover" some interesting ideas about how such a project should be designed and run, that you can share with your client, "which she might find interesting."

• Create ways to let your client have her way.

  If you succeed at driving changes in the project's requirements and your client's expectations, your client might feel like you are usurping her authority, so it could be judicious to make sure that there are aspects of the project over which she can assert expertise and control. Whatever about the project is within your client's expertise, make sure to ask her about how best to do that. In addition, as Joel Spolsky explains in "The Iceberg Secret, Revealed", graphic design is a good aspect of the project to ask your client to make command decisions about, since its perceived importance is much greater than the difficulty involved with changing it.

It sounds like your problem is a special case of a more general problem: Your client wants you to implement a system that you don't expect will work well, as specified, based on your superior knowledge of the system's domain and technology.

This question on The Workplace Stack Exchange about a graphic designer dealing tactfully with a boss who is ignorant of graphic design principles seems pretty parallel, and the answers there offer some advice that can be adapted to your situation.

• Ask lots of leading questions.

  Although your client has autocratic control of the project, its requirements, and her ultimate behavior as a moderator, you have some leverage, as the person implementing the project, to tweak the technical solution and also nudge your client's thinking toward a system behavior that will work better. You can ask questions about how she wants certain aspects of the project implemented in ways that bring up issues you are concerned about. For example:

  Should I add a filter that automatically rejects submissions that don't include a TPS number? What should it tell the user about why that number is necessary?

  How do you envision the review/improvement cycle working? Should I include a messaging system that you would use to explain what work is needed to improve a submission before you'll accept it?

What do you think of this streamlined version of the entry form?

• Introduce third-party sources.

  Your client is apparently uninterested in learning about what it's like to moderate an online forum, so it may be difficult to get her to listen to third-party sources. However, you may be able to sneak them through the door by bringing them in to help with your implementation. For example, say that you and the rest of the implementation team want to get some fresh ideas about designing a forum like this, and bring in someone with experience doing so to give a little seminar to the team, and invite your client to attend. Similarly, you could read a book or authoritative online resource, ostensibly for the purpose of getting implementation advice, and then "discover" some interesting ideas about how such a project should be designed and run, that you can share with your client, "which she might find interesting."

• Create ways to let your client have her way.

  If you succeed at driving changes in the project's requirements and your client's expectations, your client might feel like you are usurping her authority, so it could be judicious to make sure that there are aspects of the project over which she can assert expertise and control. Whatever about the project is within your client's expertise, make sure to ask her about how best to do that. In addition, as Joel Spolsky explains in "The Iceberg Secret, Revealed", graphic design is a good aspect of the project to ask your client to make command decisions about, since its perceived importance is much greater than the difficulty involved with changing it.