1) I understand the issue as follows. The Gemara (Sotah 6a) states that if there are witnesses overseas who saw her being unfaithful, the Sotah waters will not work. The Gemara learns this from the verse, "... there is no witness on her," which teaches that the water works only if there is no one who knows what happened. The Rambam (Hilchos Sotah 3:23) writes, "The water checks only a woman for whom there are no witnesses to tell about her Z'nus."

2) I understood that this means that the whole idea of Sotah water applies only when there is no better way of deciding the issue, but if there are witnesses who actually saw what happened, this is a preferable way of determining the matter, and in such a case there is no need for the Torah to give Sotah water the power to reveal the truth. Tosfos (6a, DH v'she'Ba'u) adds that it is probable that even if only one witness saw what happened, this is also sufficient to make the waters redundant. Since one witness is believed for Sotah, it follows that one witness also makes the water unnecessary and ineffective.

3) So the reason why we are not concerned that when there is one witness people will not come to doubt the effectiveness of the water, is not because the guy who saw the incident will not suspect that the waters never work, but rather the reason is because if there is a witness to what happened, we can assume that people will get to hear about this witness, and therefore they will know that the whole idea of Sotah water is not applicable in this scenario.

4) In contrast, if a Sotah is given dyed water to drink, the issue is totally different. This water will never produce a positive result. If one does such a test, which cannot possibly work, people might come to believe that Sotah water never works.

5) I found support for this in Milu'ei Even, by Rav Chaim Pinchas Scheinberg zt'l, on the Avnei Milu'im (end of Siman 178, the very end of Shulchan Aruch, Even ha'Ezer).

a) He writes something similar to what I asserted. He asks the following question on the Rambam cited above: Why does the Rambam write, "... there are no witnesses to tell about her Z'nus"? Why does he not write simply, "... there are no witnesses that she was unfaithful"?

b) Rav Scheinberg answers that the Rambam is stressing that one requires witnesses who practically can come to Beis Din and testify. It is not sufficient that the witnesses know what happened, but they must also be capable of informing us what happened. If they are capable of actually coming to Beis Din, then we say it is like they have come already. If so, the water is no longer needed. Since the water is no longer needed, the Torah did not give it the power to check the woman.

c) Rav Scheinberg concludes that it is logical, as a practical Halachah, that if there exist witnesses who know what happened, but for some reason they certainly will not come to Beis Din, it follows that they do not qualify as the overseas witnesses that the Gemara (6a) tells us prevent the water from being effective.

6) Concerning the fake water, the Rambam in Perush ha'Mishnayos (Eduyos 5:6) writes something similar to the Ra'avad. He writes that they did a trick to make people think it was Sotah water. If one allows the Sotah test to be done in such a non-genuine way, this will certainly lead to public disdain of the entire Sotah process.

7) More on why people might say the waters do not work:

Before going further, I think I should note where the Gemara states that we try to prevent opportunites for people to claim that the Sotah waters are not effective. The first Gemara is on 9b, which tells us that the Kohen warns her that first her stomach will be damaged by the bitter waters and afterwards her thigh. This is "she'Lo l'Hotzi La'az Al Mayim ha'Marim" -- in order that people not speak badly about the bitter waters if they do not know the order in which they affect the body and mistakenly think that they are not operating in the way that the Torah warned that they would. The other Gemara is on 32b, which tells us that the woman is informed of the Halachos of the waters before she is given them to drink. We tell her, for example, that if she was Shogeg, if she thought erroneously that her husband had died, the water will not check her. The Gemara says that the reason why we tell her these Halachos is "she'Lo l'Hotzi La'az Al Mayim ha'Marim." Rashi explains that if she was Shogeg, for example, and drank the waters and survived, she may come to think that even if she would have been unfaithful intentionally she also would have survived.

8) Now I want to comment on what you wrote, Elisha, that the only reason somebody might say that the water does not work is because he himself saw them having relations, but then drinking the bitter water had no bad effect on her. We now see from the above two Gemaras that there are other ways that people can make mistakes; either people see the water working in a different way than they thought the Torah said they work, or the woman herself thought the water should kill her but did not, and was unaware of a detail that made her exempt from the punishment. In fact, I can now add an additional way that people could make a mistake. This is based on the Rambam (Hilchos Sotah 3:18) who discusses the husband who gave his wife the waters to drink, even though the husband knew that he himself was guilty of having had forbidden relations with a different woman. If the husband had forbidden relations, then the Halachah is that the waters do not check his wife either. What happens when the wife drinks and survives is that she may tell other women that she was unfaithful but still managed to survive the waters. The unfaithful wife does not even know herself that the reason she survived is because of the sin of the husband, but meanwhile it is the reputation of the waters that suffer.

9) We also observe that the above cases of the waters getting a bad reputation are very different to the Ra'avad's case. The cases in the Gemara are where the waters are real, but there is a side reason for why they did not kill her -- because one of the Halachic conditions was not fulfilled. In the case of the Aruch or the Ra'avad, it was not real Sotah water that was given to her, or there was no Sotah scroll that was written for her but simply a fake piece of paper. It is obvious that such a cheapening of the Sotah process is going to lower its honor in the eyes of the people.

Kol Tuv,

Dovid Bloom