To support the problem identified earlier: "75% of all significant delays (those over an hour) are first posted by the airlines within 30 minutes of departure." Yet, in most cases you have to check in at least an hour before departure. So, this statics actually works against your service.

The point here is that we tell you 6 hours before departure--long before the airline posts anything and long before you check-in.

If you know your flight has a certain probability of being late, are you going to risk arriving late at the airport and missing the plane if and when FlightCaster gets it wrong?

Another inherent problem is related to the fact that until the airline admits that the flight is going to be late (and sometimes even after that point) they still require you to check in on-time or risk losing your seat and pay extra for rescheduling. Then, what good is the information?

As customers of the airline industry we have been trained to adjust our behavior and expectations regarding timing without being able to influence the industry. The service will NOT and cannnot change it.

Given the factors involved, it is unlikely that the model used is able to explain substantial amount of the variance. Hence, the ability to predict low probabilities.

Sorry, I will not invest a penny in a company like this.