SF is always a rare and exciting predecessor type to use in a project. For the original topic in the list, the task cannot finish until the predecessor starts. Therefore, Lag is considered overlap going forward and not backward (since you are backward passing the logic). If the "preparation work" is required prior to go-live, then no lag with whatever duration the work requires will provide the correct behavior (as you stated your desires). Negative lag goes against the "task cannot finish" logic (the true letter of intent) so I can see why Workfront prohibits the negative lag. To the thread immediately above, sequencing multiple events from an anchor point in the middle will not calculate a critical path, since everything on the path hinges from a fixed date in the middle of the project. Any "predecessors" are actually successors and therefore would not move the date. Instead, performing a forward pass of the schedule with a start no earlier than constraint (or finish no later than) on the event in the middle will show you compression to the timeline with float leading up to the event. Now, what is hard to see from the "middle" is if anything required for the event can impact anything subsequent to it. You would have to tie the predecessors in the subsequent tasks to the pre-event tasks as well as the event to show the impact to the overall project. This is one of those situations when it depends on what you are trying to have your model tell you. What dates are OK to miss: the tasks leading up to the event, the event in the middle, or the tasks after the event? When tasks are getting missed, where is the compression? Finding tasks in the middle, without a milestone view or custom field/report is hard to do since projects are really focused on a goal (end or beginning of a project in total). Just my $0.02. Dale Whitchurch Arthrex Inc