When you gamble, do you make one huge bet, or do you make many small bets?
Let's say that there's a 10% chance of your caravans getting hit, and you have 100 bushels surplus every week for 10 weeks--total of 1000 bushels.
You could send 10 caravans , and lose one.
Or, you could send 20 caravans, and maybe lose two. It's the same amount of food lost. But, maybe you'll only lose one, in which case you cut your losses in half.
Or you could send one caravan. You probably won't lose it, sure, but if you DO lose it, you'll hate yourself for it forever.
"Don't put all your eggs in one basket."
And over time what would the amount lost under both cases tend towards? If this is seriously what people are doing I would suggest we need to start charging each separate caravan sent. But you are making the mistake here of biasing the results. You could send out the ten caravans and lose anything UP to 10, the same goes for the 20 caravans. Only over a VERY large sample are the results going to result in a true 10% loss of total caravans sent.
End result, no matter HOW many caravans you send, the chances gods could end up making you pay for it. It is only OBSERVER bias that makes people think that sending the extra caravans building in protection. Thus my comment is correct. People may decide to send more caravans because they think that is a safer option, but we don't need to modify game mechanics to prevent this, because it is not a true advantage.
Also your understanding of gambling systems seems flawed. Many of the systems out there proscribe betting reasonable to large amounts, since the pay off for small amounts don't tend to cover your losses. Depending on what you are gambling on the system tends towards what your bets are, and the return odds. Its not about having so many bets that some win, since you have spent an inordinate amount on bets that don't.
