See http://en.wikipedia.org/wiki/Nuclear_weapon_design#Warhead_d...
I don't know how big the primary on that bomb was, and can't find anything about that, but did find this (http://en.wikipedia.org/wiki/Castle_Koon) test of a fusion bomb where the primary didn't fizzle but the secondary nevertheless did not ignite. The primary there was 110 kilotons. So the primary on that bomb, not fizzling, is around 5 or 6 Nagasaki blasts.
According to Wikipedia, the Mark 15 had a real thermonuclear second stage that generated significant yield. However, its second stage was also packed with enriched uranium, and it obtained most of its energy from fission.
Essentially, it was a two-stage thermonuclear weapon that had a lower fusion fraction than most two-stage weaopns.
Summary of atomic weapons:
1. Pure fission. Fusion contributes 0% of yield.
2. Boosted fission: Fusion is used solely to improve the explosive yield of the fission. Fusion contributes 1% of yield.
3. Dirty two-stage thermonuclear: Fusion contributes between 1 and 50% of yield. Most of the yield comes from fissioning of the U-238 tamper by fast neutrons.
4. Clean two-stage thermonuclear: Fusion contributes 90-99% of yield. The U-238 tamper is replaced with beryllium or lead, which does not undergo fission when hit with fast neutrons.
5. Three-stage thermonuclear: Same as two-stage. Thermonuclear bombs are considered to be an "unlimited" design, because you can keep stacking stages on top of each other, until you've got enough to blow up the planet.
A fizzle in the Tsar bomba (a 100MT capable device) would not be expected to release 1MT of energy or even 1/10th that, it would still release likely very much less than even a single kiloton.
Should assumes particular goals, and with almost any system you're looking at a trade off between safety and reliability. Like the sensors on a modern car; some cars refuse to start these days if the sensor's out.
The military's perspective is likely to be something like: We press a button, we want a mushroom cloud. Thank you very much. They're unlikely to want to introduce unnecessary points of failure in their system.
Timing systems, trigger mechanisms and the like will break, or be manufactured incorrectly. Nothing's perfect. Especially when you're designing something that's going to be thrown around a bit (slimmer margins.)
Which then gets weighed against the concern that that thing better not go off by itself.
I would imagine that the latter is why there were multiple safety devices on the weapon - reasoning that the chance of them all failing concurrently was very low.
Either way you do it, you're taking on some degree of risk of something - even if you make a system with a single point of failure, and make it fail safe as far as possible, that can never be entirely eliminated.
It seems rational to us that you'd err on the side of the thing not going off. But then again we're not in an arms race right now. Someone might well reason that if you introduce something that's going to introduce a certain failure rate into your system, that's going to mean you have to invest in more weapons to reach the degree of force you want. Especially if they're reasoning in a very constrained space - i.e. they primarily think about the threats they're used to - and just don't think of things like bombers going down, (which in any case they're going to find themselves encouraged not to do since it implies fault on their part) that's going to be a very easy rationalisation for them to make to themselves.