I believe the only users that should be impacted would be users with the Customize Application permission who continue to make modifications to your org's configuration during the deployment validation process (such as modifying a validation rule or updating a sharing setting). They should get a message that the org is locked and be unable to save their changes, which is more of a disruption in their access than anything that could impact the results of the validation. However, it is possible that a user could be in the middle of some long-running process (such as running test cases or updating sharing settings) that could prevent the validation from even starting in the first place.
Also, if your test cases rely on certain data being present and in a particular configuration (rather than mocking up the data as part of the test case, which is a best practice), allowing users to make changes to data in the org during the validation window could potentially impact the success of your test cases. Of course, this is easily preventable by writing test cases that create their own data or ensuring that they utilize the
@IsTest(SeeAllData=false) annotation introduced in API version 24.0.
Other than that, your end-users really should not be impacted by a validation-only deployment, as all changes introduced are never actually committed and will be rolled back following the completion of all tests. In my experience, locking out all users should only be necessary during the actual deployment window (especially when manual configuration changes are necessary in addition to the automated deployment and it would be undesirable/dangerous for a user to be accessing the system prior to the completion of the manual changes).