In my experience, when bids take a jump, it is a quality score issue. Your keywords aren't relevant, or specific enough to your website. Tighten up your ad and landing page around the keyword(s) you are having trouble with. Make sure the keywords really are relevant to your website.
Your bids shouldn't be based on a pre-determined strategy. You should be setting and adjusting bids based on their conversion rate. The higher the conversion rate for a keyword, the more you can afford to pay for each click. For example, if keyword A has a conversion rate of 5% and keyword B has a conversion rate of 1%, then bidding £0.50 for A and £0.10 for B will both deliver conversions for £10 each. Whether you should start with a large number of terms and whittle them down, or a small number of terms, and add new ones over time, is quite a tricky question to answer. There's no point in having a huge number of terms if each will have very low volumes of traffic - you'll never be able to tell whether they are working or not. So it could be argued that if you have a small budget, then a small keyword list is a good idea. On the other hand, the 'long-tail' of very specific keywords tend to perform better, so there's an argument that you should be focusing on these if you have limited cash. My approach is to look at the long list, and decide whether grouping them into good Adgroups will allow me to analyse them by group. If you have a lot of keywords that are variations on each other, this is usually possible, so you can bid on the long-tail without being frozen by a lack of data. If, on the other hand, you have a lot of keywords that aren't really groupable in any meaningful way, then you may have problems optimising your campaign.
I had a general concept of how Google determines the price of my keywords and your post is informative and reasonable in its assertions. I always allow myself a cooling off period before diving into a literary tirade in the blogosphere, and yes, I realise a huge, wealthy corporation could care less what I think about them as long as their revenue stream has not been impacted. I did notice that one of the moderators deleted my reference to Brin and page, suggesting that Euromania was good for business. Still, I feel 5.00 USD is an outrageous price for a keyword. And who knows how high the price might have gone had I not discontinued my campaign? Once the next billing cycle has passed, and Google has extracted its graft from my bank account, I will be canceling my account and investigating the services of Google's competitors. I hope Sergei and Larry apply my funds towards the purchase of a conscience. B.