This thread is only for testing the new Sentiment Analysis process. The former one appeared to assign values that were different from a human reviewer more often than I would like. So far the new version is living up to expectations. Though not infallible, it appears to arrive at non-intuitive results much less often than its predecessor. This particular post should be rated neutral or close to it.