Reason dictates we assume the negative when there is not conclusive evidence towards the positive claim.
We also have to consider the risks of the various results, and weight them against our priors.
If we think something is fairly unlikely (1% chance of being true), and as yet has no evidence to support or refute it, but would cause the total destruction of the earth, reason would dictate that we assume that it is true until we have evidence that it is not.
This is obviously quite sensitive to what our individual priors are about something in a world where we don't yet have any evidence. This will naturally produce different conclusions for different people (you may assess the world-destroying event to only have a .000000001% chance of occurring, in which case it's pretty reasonable for you to assume it's false until we have evidence that it's true). But, for any individual, I don't believe the situation should not be as simple as your suggestion.
Edit: of course, this presumes self-preservation as a goal which is not something which can be determined by reason. Of course, it's a goal shared by a vast majority of humans, so I feel it's a reasonable assumption here. As reason/logic are used as a tool to support our goals, I feel it's fair to include such a goal as part of our reasoning consideration.
But if you choose the sensitive approach and expect evidence from the party who proposed the theory then starting from the article's premise that there's no reliable evidence (if true) is a perfectly valid argument to support the title.
Consider this: I am the Always-Right-All-Knower-and-All-Seer. Now what is more reasonable, to expect evidence from my side supporting my premise, or for you to provide evidence against it and assume the premise true until you do that?
> I am the All-Knower-and-All-Seer.
My prior that this is true is extraordinarily low. I put this at a 0.000000001% chance of likely to be true. Suppose you provide me a theory that Iraq possess Weapons of Mass Destruction, and must be invaded in order to prevent the destruction of a major city.
I would assign an extremely-high negative consequence to that outcome. I would assign a low-ish probability to the outcome based on my inherit disbelief that you are Always-Right-All-Knower-and-All-Seer, and that you have not furnished any evidence.
If your proposed solution is to invade the country, I would assign a very-high negative consequence to the outcome. As such, I would ask you to provide quite compelling evidence that your claim was true, before supporting such a rash action.
If, instead, your proposed solution was to propose international monitoring to the country of Iraq I would assign a quite-low negative consequence to the outcome (probably even a slightly positive consequence). As such, before backing your proposed solution I would require a much lower bar of evidence.
But I believe you about the WMDs in either scenario to the same degree. Your proposal changes how much I need to accept your premise about the WMDs to accept your solution.
So, I think accepting decision D based on belief X is a far far more complicated question that "do you believe X"?
----
Let me also look at this from the other side. Let's say I, for some reason, conclusively believe you to be an All-Knower-and-All-Seer. My belief in this is 100%. Perhaps I am easily convinced, or perhaps you made 1,000,000,000,000 predictions of fully stochastic events, and I have assigned a 10-sigma confidence that you are some kind of All-Knower-and-All-Seer. Let's also assume that I have 100% confidence that you are honest, and that I have some way of gaining that confidence. However I got there, it's somewhat irrelevant at this point.
If I have 100% confidence that you are truly an All-Knower-and-All-Seer, and you suggested a particular course of action was strictly beneficial to me, I would almost certainly follow that course of action without requiring further evidence from you.
I would argue that, in such a scenario (and such a scenario is quite contrived, I believe, for anyone who is truly skeptical) it would be irrational for me to ask you for more evidence before acting. After all, at that point we've already moved by belief threshold high enough for me to take actions in the world to further my goals.
This all assumes that we are an actor in the world with goals (which, I again assert that goals cannot be defined rationally, but rather that rationality serves as a tool to achieve goals). If you have no goals, then I agree that pure reason would require you to ask for more evidence.
----
My return hypothetical is to ask you this question:
You are a key decision maker for a scientific body. You are deciding whether you should run an experiment code-named Alpha. There is a low-likelihood but not preposterous theory (say, accepted by 10-15% of the scientific community), but lacks any solid experimental support. This theory says that running the Alpha will destroy the planet. The gain from running the experiment will be furthering scientific knowledge in quite useful, but not ground-breaking ways.
Again, this "Destruction Hypothesis" seems like it's not super likely to be correct, but it's not a crackpot-theory.
Do you run Alpha today without delay, or do you order more investigation of the Destruction Hypothesis to conclusively rule it out before running Alpha?
You're free to believe an unsubstantiated theory but you can't expect others to prove it wrong in order to make their point.
Now coming back to your hypothesis, it's too complex and contains way too little data to assist in a decision. A random answer would be just as good and proves nothing. I would probably consider performing the experiment on another planet. Or even on this one. Or wait a little longer (which is equivalent to not doing it). What I would definitely not do is say "The premise of your entire [theory] seems to rely on this one fallacy" [approximate quote from OP]. It's only a fallacy after you prove it.
P.S. Is experiment Alpha the detonation of a thermonuclear bomb that might set the atmosphere on fire? Because we know how that decision went :).
You have a decision to make, and Bob makes a claim which could influence that decision.
Does it matter where the burden lies? You have to make the decision, and Bob has presented new information but doesn't well substantiate it.
My point is this: From a pure reason standpoint, I think it's fair to always assume the negative when there is not evidence for a positive claim. But in the real world where you have to achieve real goals based on partial knowledge, reason would dictate that you sometimes evaluate and consider claims that are not yet well substantiated, and you might have to do the legwork to substantiate the claim yourself (if you decide there is enough merit and value in doing that substantiation).
My point is the burden of proof is, practically, often on the person who needs to make a decision based on information and not on the person who makes a claim with related information.
---
For Alpha, I didn't have something specific in mind but some kind of combination of "light the atmosphere on fire" and more crackpot "your particle accelerator will create a black-hole" were definitely both in the back of my mind.
> the burden of proof is, practically, often on the person who needs to make a decision
You're contradicting a logical principle that is strong enough to be put in actual law.
It's the difference between being right and pretending or hoping you're right. Or hoping that the other guy is wrong.
> The premise of the entire article seems to rely on what might be a fallacy.
Just doesn't have the same righteous ring to it :).