I have an MBA from a school that sometimes makes the Top 50 lists. I had some extra scholarship semesters available as an undergrad and was allowed to partake in the MBA program concurrently.
The MBA experience shaped the way I think about problems as much as my undergrad in computer engineering. A good program teaches students to think through problems using an economic framework while engineering school promotes a technical one. Both are valuable (as are many other ways of thinking).
Here's a specific example: Is Amazon's new EC2 dedicated instance pricing a good deal? A naive analysis requires simply adding up the sum total of all costs and comparing the difference. What's wrong with that?
1. A dollar today is worth more than a dollar tomorrow. How much more? That depends on your unique circumstances. You're a start-up? It's really expensive to spend money today. You'd much rather pay more later when your concept is proven out. Are you GE and can borrow at rates only slightly higher than the government? Then, a dollar now isn't much different than a dollar tomorrow. So, you have to establish what's referred to as a discount factor to compare payments made at different times.
2. The dedicated instance program has characteristics of an option. You have the right but not the obligation to purchase server hours. So, your economic model has to allow for the future decision to switch over to the per-hour model if prices drop drastically.
3. The "strike price" (using the term loosely) of the option is fixed even though Moore's law is likely to hold over the life of the option. Is paying $500 now for the right to purchase server hours for 70% off a good deal if server hours only cost 1/5th of what they do now at the time of the option's expiration? How steeply can prices decline so that you are still at the point of indifference between paying up-front and paying later?
4. A quick analysis of the marketplace for on-demand computing suggests that it is quickly becoming commoditized (at least at the basic levels). Price competition for commodity-like goods is tremendous, so we should expect prices to drop as more competitors emerge. Even if Moore's Law didn't hold, one shouldn't evaluate the option based on today's pricing.
5. The option doesn't appear to be transferable. What discount factor does one place on purchasing something valuable that he can not sell?
I actually put together such a model. Even under optimistic Moore's Law cases and high discount rates, dedicated instances are a good deal (presuming you will use all three years with certainty or can transfer your rights).
The MBA experience shaped the way I think about problems as much as my undergrad in computer engineering. A good program teaches students to think through problems using an economic framework while engineering school promotes a technical one. Both are valuable (as are many other ways of thinking).
Here's a specific example: Is Amazon's new EC2 dedicated instance pricing a good deal? A naive analysis requires simply adding up the sum total of all costs and comparing the difference. What's wrong with that?
1. A dollar today is worth more than a dollar tomorrow. How much more? That depends on your unique circumstances. You're a start-up? It's really expensive to spend money today. You'd much rather pay more later when your concept is proven out. Are you GE and can borrow at rates only slightly higher than the government? Then, a dollar now isn't much different than a dollar tomorrow. So, you have to establish what's referred to as a discount factor to compare payments made at different times.
2. The dedicated instance program has characteristics of an option. You have the right but not the obligation to purchase server hours. So, your economic model has to allow for the future decision to switch over to the per-hour model if prices drop drastically.
3. The "strike price" (using the term loosely) of the option is fixed even though Moore's law is likely to hold over the life of the option. Is paying $500 now for the right to purchase server hours for 70% off a good deal if server hours only cost 1/5th of what they do now at the time of the option's expiration? How steeply can prices decline so that you are still at the point of indifference between paying up-front and paying later?
4. A quick analysis of the marketplace for on-demand computing suggests that it is quickly becoming commoditized (at least at the basic levels). Price competition for commodity-like goods is tremendous, so we should expect prices to drop as more competitors emerge. Even if Moore's Law didn't hold, one shouldn't evaluate the option based on today's pricing.
5. The option doesn't appear to be transferable. What discount factor does one place on purchasing something valuable that he can not sell?
I actually put together such a model. Even under optimistic Moore's Law cases and high discount rates, dedicated instances are a good deal (presuming you will use all three years with certainty or can transfer your rights).