You guys are doing the math wrong.
If you have two things with a 10% chance of failure, the chance of both of them failing at the same time (assuming they are completely independent) is 1%. But the chance of one or the other of them failing is ADDITIVE, because they are totally independent. (See
Addition Rules for Probability) In perfect math, the chance of one of the connections failing is actually slightly smaller, because you have to subtract the probability of BOTH of them failing (unless you include that as a subset of one connection failing). But it's additive, not mutiplied, and the more possible failure points you put into a system, the greater the chance that some one of them will fail.
However, in recreational diving, where the surface is always an option, a leak from a right angle adapter is unlikely to be disastrous. AJ and I inhabit a diving world where even small failures can be of tremendous impact -- a big leak from a connector can be a major problem, if you have a couple of hours yet to swim in order to end your dive. To cope with that, we use a system which has MANY more failure points than a single tank setup, and which requires the diver to learn a whole set of behaviors to cope with those failures. We put up with that because what we get in return is redundancy, so that in the case of the vast majority of common failures, we cannot be left without a gas supply.
Any decision one makes about gear has to consider the benefit that a change offers, and the potential downsides of that change. Adding another static o-ring connection is a pretty low risk change.
This is on the mark.
The type of failure, or the result of that failure are also important. For example, second stages tend to fail much more often in a freeflow mode than they do in a failure to deliver gas mode. About the only way for a second stage to stop delivering gas is if the poppet slips past the feet on the lever. If the reg is properly maintained and the arms on the lever are not worn, it's a very low probability event. The other "no gas" failure mode is for something to block the lever and prevent it from opening the valve. Unless you are in an area where gravel may enter the second stage, that's a very low probability event as well.
One thing you see no and then is someone with 3 second stages (primary, octo and Air 2, for example). The odds of a failure to deliver gas are as noted incredibly low, So for the 3rd second stage to be useful, you'd have to have 2 second stages suffer the same very low probability failure, or alternatively have a second stage suffer a very low probability failure and have a team mate run out of gas - a should be low probability. However to gain this dubious benefit the diver now has three second stages that could each individually suffer the higher probability freeflow type failure - and any one of those regs suffering that failure causes loss of all the gas in the system. So if the probability is only .01 of a single second stage failing, then with 2 second stages, it's .01+.01=.02, and if you add a third second stage to the same system, it's now .03 - a 50% increase of one of the second stages failing and causing loss of all your gas.
The point then is that a single tank diver does not carry an alternate second stage to bail out to if his primary fails (as a failure to deliver gas event is very, very, low probability, and if it did happen, you bail out to the buddy's octo). Instead the alternate second stage is there for a buddy to bail out to if he loses all his gas or access to his gas (covering the higher probability forgot to monitor the gas and ran out failure, but also the very low probability failure to deliver gas second stage failure in the team).
This approach of adding a second stage and greater potential for gas loss was chose as buddy breathing was considered difficult for divers to do effectively. The cost of the approach however is that the probability of gas loss through a malfunctioning second stage
doubles for the 2 second stage diver compared to the 1 second stage diver. Now, when you think about it, there's no way you'd want to make it even worse by adding a third second stage as it adds no benefit but 50% more risk of a second stage failure. However people do exactly that because they don't understand the
purpose of the equipment in mitigating risk, nor do they understand the risks that additional equipment both mitigates and
adds.
And once again, adding a right angle fitting does add an additional o-ring, and that additional o-ring has a finite probability of failure. With good maintenance practices and tool-tight fittings, that probability is very low. Significantly, the most common failure mode is a champagne style leak that won't cost more than a breath or so over the course of the dive, so even if it does occur, it's a non-lethal problem.
One of the ironic things I see some technical divers do is leave the inlet fitting hand tight. The idea here is that in the event of a very low probability "failure to deliver gas" second stage failure, they can swap second stages. The problem with this brainstorm is that the hand tight fitting now has a fairly high probability of coming loose enough during a dive to blow out the o-ring. When that happens it blows out about a third of the o-ring and you get a very significant leak, so even feathering the valve loses a lot of gas. Plus you've blown the o-ring in the hose fitting, so swapping second stages won't fix the leak. But again, divers do exactly this as they fail to understand the actual risks and the increased risk of their flawed mitigation strategy. (If a diver really wants this second stage swap capability, they need to leave the fittings tool-tight and keep a wrench in a pocket.)
You're simply not paying attention to the variables, just the numbers. Gotta watch both. Adding more gear increases your chances of something breaking/failing. I've done more than a handful of dives where there were about 30 (thirty) regulators within the team, not counting the setup guys. 30. You're nuts if you think that having thirty regs decreases the chances of ONE of them failing.
And I think something has been lost in translation over the interwebs regarding the swivels. You mentioned something regarding confusing redundancy with failure points. I am not at all confused on the two. Redundancy gives you options when something fails at the expense of added chance of someTHING failing. I can get down with that. Swivels don't add much to most people's diving kit, yet increase orings and moving parts. I don't get down with that.
And the above considerations hopefully illustrate why AJ is correct in the importance of not just the math, but also the actual variables and the failure modes.