If determining a date where simply a matter of determining the average of the possibilities than a "middle" date (i.e. one from c. 70-100) would be preferred. But that is not adequate historiography. A more robust procedure would inquire into whether there is evidence for whether John should be located before the 70 divide or after. Let us begin by considering the evidence for a post-70 date, beginning with 4:21. When Jesus tells the Samaritan woman that "a time is coming when you will worship the Father neither on this mountain nor in Jerusalem" must he not surely be referring to the destruction of the temple in 70? Two observations are in order. One, it's reasonably clear that the point of 4:21-24 is not the destruction of the temple but rather the soteriological consequences of Jesus' ministry. Two, that the statement is a prophecy after the fact referring to the destruction of the temple during the Jewish War seems highly unlikely given that it is mated with a prophecy that worship would also cease on Mt. Gerizim...something that did not happen during the war. A prophecy after the fact that reports blatant non-facts seems curiously strange. 4:21 probably tells us nothing about the date of the gospel. Similarly, 2:19-22 probably tells us nothing either. Yes, Jesus is reported to have uttered a prophecy that the temple would be destroyed and rebuilt, and yes, we read that the disciples eventually interpreted that as a reference to his own body. That reinterpretation could have occurred after the destruction in 70, in order to explain why the temple was not rebuilt, but it could have just as easily have occurred after the resurrection: exactly as 2:22 states. Indeed, I see no reason to doubt John on this matter, which suggests that the reported reinterpretation could have occurred as early as 30. Again, in turns of defining the date of John's Gospel, 2:19-22 is probably non-probative.
Sometimes it is argued that the Gospel must post-date 70 because 9:22, 12:42, and 16:2 clearly suppose the existence of the rabbinic prayer Birkat ha-Minim, which dates to the post-70 period. This is a non-starter, on so many levels. In fact, it is a non-starter on so many levels that I've written an entire book on the matter. The argument ignores a series of facts. For instance, we can be more confident that John's Gospel dates to the first century than we can be that the Birkat ha-Minim does. Or, we cannot be confident that it was originally an anti-Christian prayer, as this reading of 9:22, 12:42, and 16:2 must suppose. Or, 9:22, 12:42, 16:2 refer to expulsion from the synagogue with no references to prayer, while the material surrounding the Birkat ha-Minim refer to prayer with no references to expulsion from the synagogue. We are left trying to offer as background to 9:22, 12:42, and 16:2 by invoking a text that more likely post-dates John's Gospel than not, probably did not have the aim that it must have had if it is to speak to these Johannine passages, and in fact lacks almost any actual parallel with John 9:22, 12:42, and 16:2. New Testament scholarship's fascination with the Birkat ha-Minim is mischief best laid to rest.
In sum, I see nothing in John's Gospel that should incline us towards a post-70 date. Is there data that should incline us towards a pre-70 one? I think that there is. A key datum is 5:2, which tells us that "Now in Jerusalem by the Sheep Gate there is a pool, called in Hebrew Beth-zatha, which has five porticoes." Note the present tenses here. This is exceptional for John. Typically, he assimilates his topographical descriptions to the narrative, typically using the imperfect. The slippage to present tense makes best sense to me if the author most spontaneously thinks about the Sheep Gate as a present reality, and that such thought has led him to neglect assimilating this description to the time of the narrative. Such spontaneity makes far better sense before 70 than after.
One might object that 5:2 cannot bear this much weight, and of course in a fuller treatment I would furnish more data than just this, but I would suggest that if there are no data that points at a post-70 date but there is a datum that seems more "at home" prior to 70, then pre-70 is the safer bet.