Regardless, I think fixed query plans in SQL are a very bad idea and at odds with SQL's declarative nature.
By trying to mitigate against one time bomb by fixing the query plan, you introduce another time bomb where that fixed query plan, which performed well on all your tests, performs horribly with certain parameters (e.g., parameters that cause one side of a JOIN to be much less selective than the query plan had allowed for).
Once you discover some input on which your query performs horribly, you need to come up with a new query and you're going down the rabbit hole of trying to out plan the DB's own query planner because you wanted to save on some DBA effort.
If you're encountering sudden thousand-fold slowdowns there are probably issues with your database design and you'd do well to investigate them rather than fighting against the DB trying to keep everything fixed so it doesn't break.
Out of curiosity, do you have any specific examples of these slowdowns which are simple enough to describe here?
Hardware is "hard", but it's hard for the competition too - if you're good in that area that should be what you want. The social network / web app market has few barriers to entry - the programming is not difficult enough to prevent average programmers with great marketing and design skills from entering. Not good if your strongest asset is your technical ability and you hope go head-to-head with web 2.0 hipsters on there terms.
If you can handle something that requires math, physics, hardware knowledge, big initial investment, etc, why not look into ways of monetizing it? It's tough but you'd have a smaller pool of competitors than if you decided to try and make the next Flcker/Facebook/Twitter/etc. That more people aren't doing this leads me to believe I'm overlooking something, but it's surprising so many geeks are drawn to (and succeeding at) startups that seem to be the antithesis of geekiness (social networks, advertising, etc).
In complexity theory, n is based on the length of the binary encoding of the input, NOT the decimal value as was used here. Therefore, the x-axis should be based on n = log2(k), for each number k.
Overlooking this would lead one to believe there are known algorithms that solve NP-C problems in polynomial time (e.g., knapsack problem can be solved in polynomial time with respect to the decimal values of its inputs, but exponential with respect to the length of the binary encoding).
I would love to see a study done comparing the benefits of Ritalin, Adderall etc. to users diagnosed as having ADD/ADHD against users not diagnosed as having it. If the former group receives significantly greater benefits that would be a strong indicator that ADHD is a distinct condition, rather than a label applied to a set of personality traits. On the other hand, if the benefits were similar for both groups it would call into question why one of them is denied (legal) access to the drug.
According to the Time Hierarchy theorem ( http://en.wikipedia.org/wiki/Time_hierarchy_theorem ), for each number k > 1, there exists a problem in P that can be decided (solved) in O(n^k), and not decided (solved) in O(n^j) for each j in [1, k).
This was the first time I heard of this theorem (was googling for something I thought would weaken the author's point), and it really strengthens the authors argument - I thought his example of O(n^10) algorithms was sort of a fallacy but it appears not. If I'm properly understanding the Time Hierarchy Theorem, it is a very strong argument for what the author is suggesting (and somewhat disappointing for me, because I wish he were wrong).
I don't usually enjoy any of the activities I suggest as ideas for "dates" in and of themselves (on the occasion I need to come up with one). I just try and think of something that is cheap, but will not seem like I chose it just because it's cheap.
Do you think that people's "what-to-do" dating ideas are an accurate reflection of who they are?
SQL queries inside a PL/pgSQL function are also often but not always cached (don't know the exact rules): http://www.postgresql.org/docs/9.4/static/plpgsql-implementa...
Regardless, I think fixed query plans in SQL are a very bad idea and at odds with SQL's declarative nature.
By trying to mitigate against one time bomb by fixing the query plan, you introduce another time bomb where that fixed query plan, which performed well on all your tests, performs horribly with certain parameters (e.g., parameters that cause one side of a JOIN to be much less selective than the query plan had allowed for).
Once you discover some input on which your query performs horribly, you need to come up with a new query and you're going down the rabbit hole of trying to out plan the DB's own query planner because you wanted to save on some DBA effort.
If you're encountering sudden thousand-fold slowdowns there are probably issues with your database design and you'd do well to investigate them rather than fighting against the DB trying to keep everything fixed so it doesn't break.
Out of curiosity, do you have any specific examples of these slowdowns which are simple enough to describe here?