Now that the report's out, a couple of things need to be clarified. Please bear with a little too much explanation:
1. The interlocking had nothing to do with the derailment. The point of derailment (POD) was not at the switch, but about 200 feet beyond the nearest turnout in a curve. The train was making a reversing move from track 2 to track 1 when it derailed. If you look at the TGV exception report on page 49 of the report, you can see the line items of Frog and Switch Point in the report above (or in the test run, before) the wide gauge line item two lines from the bottom, which is near the POD. The lead cars of the train were in the middle of the interlocking, with the rest of the train on track 2 behind the interlocking. The train had to clear the interlocking on track 2 before making the move through the interlocking to track 1. If the train derailed in the interlocking, the reverse move would not have been possible because the train would not have been able to clear the interlocking with 5 derailed axles.
2.Looking at that same Exception report on page 49, you'll notice that two lines above the wide gauge defect is a gauge change defect of 1.5 inches over 26 feet at the same location.This itself is a black condition, and could be very well indicative of poor anchoring of one of the rails. The fact that the number is positive means that the gauge widened 1.5 inches in the direction the TGV was traveling, which was the direction of prevailing traffic (inbound,with the chain marker decreasing). A train traveling in the opposite direction would experience the opposite effect, with the gauge tightening a large amount over a short distance. This could produce a squeezing effect on the wheels and a subsequent derailment. This can also explain why one month of regular traffic went over this area without incident, but why one reversing move would have such devastating consequences. I'm pretty sure this defect was not deleted, which means that there should be record of the follow-up inspection of this defect and actions taken to bring the defect to at least a yellow condition. Theoretically, if the gauge change issue was resolved, the wide gauge issue should have also been resolved.
3. The TGV determines parameters like gauge by using lasers. In an interlocking, varying components such as the closure rails and the frogs produce anomalous readings, such as 59 inch gauge readings. However, these readings are usually over a very short distance and register as a "spike" on the readouts. EDIT: That are what these "false positive" statements are based on. Items like these would require a very complex software to identify, and the accuracy rate for this kind of detection would not be worth the cost of such development. Typically, items such as switch points, frogs, and platforms are manually input into the TGV during the inspection runs.
4. The gauge varies in different locations. In turnouts, the standard gauge is supposed to be 56.5". In tight curves, the gauge is widened slightly to reduce wear and the potential for a wheel climb. As was mentioned though, for tight curves like the one at the derailment location, anything over 57 inches is a slow speed order condition.
Personally, I believe the "wide gauge" and the accidental deletion explanation was a simple way to put a nice cover on a much deeper, systematic issue, as well as to provide an easy target to point a finger at. Realize that one of the recommendations was to give the TGV crew the authority to take tracks out of service if need be. The TGV crew should have had that power in the first place! The question becomes this: why, until now, did one of Metro's most sophisticated vehicles not have some sort of QA program in place, and why this program was given such little latitude to act on such bad findings?