The way he explains it, it just does a dil check on one cell, and an O2 check on another, against an expected PO2 reading for each. If they are what it assumes, then it's fine. The problem is when one fails and then it doesn't have the check against the other to validate functionality. They are not compared in the traditional sense of voting logic. At least that's my understanding. In the event of a cell failure, you are using one cell to try and maintain a loop. If it's the O2 cell that goes, it pegs you to a 1.0 PO2 and uses the dil cell to try and maintain that. If the dil cell goes, it uses the O2 cell to maintain a breathable loop. The issue lies when there is a failure of one cell or the other, in which you have zero way of validating the loop viability. Without validation, bailout is the only option. However, bailout is not always the only, or preferable option.
It's not a function of "more is better," it's a function of "more data is better." Voting logic in and of itself is subject to voting out a good cell as mentioned by @Superlyte27 however, having the available data from multiple sources allows you to determine which cell is valid. Let's say you have 2 current limited cells and one fully functional cell. You can determine which of the 3 cells are bad, and depending on where it's limited, you may be able to maintain all functionality at a lower PO2 level. You may also fly it with one cell, in which case you're in the same boat as the Poseidon. The point is, it gives you data with which to make choices. 2 cells doesn't give you a check against the other. 3 cells allows you to verify all 3 for functionality. Of course, worst case you could have 3 bad cells, but that's pretty darn unlikely.
And again, we're talking about a commercial system, not an industrial system that's been vetted by thousands of hours of unmanned and manned testing before being released into the wild. I guarantee that even one system part on a 737 has more testing before being integrated into the product line than every part of the Poseidon VI/Se7en combined.
Sure it's wiki, but it's a decent explanation on the benefits of actual redundancy and voting logic Redundancy (engineering) - Wikipedia
This talks about datacenter down time, but replace "data center" with "O2 cells" and it makes a lot of sense. Redundancy N+1, N+2 vs. 2N vs. 2N+1
It's not a function of "more is better," it's a function of "more data is better." Voting logic in and of itself is subject to voting out a good cell as mentioned by @Superlyte27 however, having the available data from multiple sources allows you to determine which cell is valid. Let's say you have 2 current limited cells and one fully functional cell. You can determine which of the 3 cells are bad, and depending on where it's limited, you may be able to maintain all functionality at a lower PO2 level. You may also fly it with one cell, in which case you're in the same boat as the Poseidon. The point is, it gives you data with which to make choices. 2 cells doesn't give you a check against the other. 3 cells allows you to verify all 3 for functionality. Of course, worst case you could have 3 bad cells, but that's pretty darn unlikely.
And again, we're talking about a commercial system, not an industrial system that's been vetted by thousands of hours of unmanned and manned testing before being released into the wild. I guarantee that even one system part on a 737 has more testing before being integrated into the product line than every part of the Poseidon VI/Se7en combined.
Sure it's wiki, but it's a decent explanation on the benefits of actual redundancy and voting logic Redundancy (engineering) - Wikipedia
This talks about datacenter down time, but replace "data center" with "O2 cells" and it makes a lot of sense. Redundancy N+1, N+2 vs. 2N vs. 2N+1