>>Basically, there are two ways in which software (or anything else) can be made more reliable - quality, or redundancy.<<<br><br>Right - parallel redundancy increases reliability. Your example of diverse fire starting means is that sort of thing. Reliability theory at its root is pretty common sense (and calculating the specifics is not all that onerous, either).<br><br>But your stove failure intrigues me. I've used a few different white gas appliances at colder temps than that, including three different Colemans (two burner green box, old style "Sportster" on a lantern body, and original Peak 1 - plus Svea 123 and MSR XGK). Fuel makes a difference, or at least it did those years for me. I used Blazo (Chevron brand, IIRC), and it was probably refined locally (Alaska or possibly Canada). In cold temps, I found Coleman fuel to be "lazy" at igniting and Blazo to be "vigorous" - too much so in hot temps (summertime) actually - at least for my comfort level.<br><br>As for the stoves - In extreme cold the Svea had to be pump primed and babied - NO wind or it would slowly fade, and always insulated from the ground. The rest worked fine. The two burner could be a beast to get buring clean - I found that some careful pre-heating of the generator helped - but once it was going, it was fine (and a new generator sometimes works a miracle). The rest of the stoves were fine - preheating was/is part of the ritual for me with those stoves most of the time unless it's warm out.<br><br>The coldest I am absolutely certain that I have used Coleman fuel in a two burner green box is -25 F. It worked fine; fuel was fresh and the stove was nearly new, FWIW.<br><br>So... maybe it was your fuel and/or an old generator? Fresh Blazo is the best I've used in extreme temps. It was available in either 5 gallon cans or 55 gallon drums and I suppose it still is. Might be worth scrounging some up and trying it out this winter.<br><br>Regards,<br><br>Tom