Stii ce vrei ?

Found this The Last Psychiatrist site via this post: When Was The Last Time You Got Your Ass Kicked?.

Found this: The talk miner.

And I got an idea that seems interesting enough to pursue a little .

Also my little sister was on tv: http://blog.yau.ro/2010/08/yau-la-stirile-pro-tv/ (which is always nice)

Ps,
And saw the video of Rehab by The Jolly Boys (Jamaica):

Bookmark It

Very funny joke from from Omid Djalili.

Later edit: Fixed video link.

Bookmark It

The P != NP proof looks like is not happening. There was a lot of discussion and formulations of the issues to the proof but for now it looks like is not the proof. Too bad . While i followed the  discussion these couple of days i saw a couple of descriptions of the problem in layman terms.

Now it’s true that every time you describe a problem for someone who doesn’t have the necessary common vocabulary to follow it’s formal description you will be inaccurate. Actually that’s the point  when you do this, you want create a simpler model of the problem using whatever mental constructs you assume / know that the receiving party has. And you hope that the main points of the problem will translate into an intuition for the receiver. It’s hard to do this properly because humans by nature are mostly self centered and egoistic, and they tend to look at the world via their internal model. That’s also the reason you rarely find excellent teachers. The be a good teacher you need to be able to make the students reach conclusions intuitively before they are able to justify them formally. You need to actually understand how they see and look at the world. But i digress .

I had an epiphany today. While reading my blogs and browsing i found an very interesting website (don’t worry i’ll post it at the end). In the moment i saw it i knew that it’s very useful to a certain demographic and has the potential to be a very big player in said demografic. The funny thing is that i know people which have been quite close to this idea but since they were limited by their world view couldn’t jump the boat and see it. The same thing actually happens in a lot of places. In entrepreneurship, science, sports you will always find people who are experts in a certain area but they usually fail to make a leap (by changing the world view) and change the game.

If we are talking about the technology for example  … when Apple launched the iPod the market was actually filled with mp3 players and they where mainly competing on price. Jobs knew (or deluded himself ) that he could make it better and change the game (don’t compete on price but compete on feeling) and he did it. In search when Google was born the market was full of search engines  who sold ads and actually Yahoo had the opportunity to build the technology which allowed Google to be so profitable (Paul Graham said so). All the pieces (the big index, the keyword bidding technology) were there a couple of years before Google was born and they didn’t do it. It took a couple of outsiders (2 students) to put the pieces together and change the status quo.

In math a lot of interesting breakthroughs were usually created by outsiders (they were outsiders when they did it): Isaac Newton (he practically invented the scientific method and solved most of the open problems of the time), Abel (he died before reaching 25 but invented Group Theory), Gauss (he rebuilt the algebra theory by age 25).

But what’s the connection with P vs NP ?. In a moment . First i will say that the following is the simplest and most intuitive description of the P vs NP problem that i saw until now:

There are problems (in computing) that are much harder to solve exactly than it is to test if a certain solution is correct ?

Basically P is a class of problems that are called of Polynomial complexity and NP is a class of problems of Nondeterministic Polynomial complexity. A problem known to be in P is the key search (finding the position of a specific key in a list of n keys ). A problem known to be in NP is The Subset Sum Problem (Given a set of integers, does some nonempty subset of them sum to 0). In order to solve the binary search you only need to execute maximum n steps (check each key and when you find the one you need you solved). On the other hand the Subset sum is more complicated to solve (the number of possible subsets for n numbers is 2ⁿ which is extremely big for even smaller values of n (for example if n = 100 then 2ⁿ will be aprox 41,213,151,536,758,394,503 years assuming that testing each variant will take 1 / 1000 seconds .. the universe is significantly younger than that). But if we are given the solution is trivial to test it’s correctness: sum the numbers and if the sum is 0 then we are in business.

The solutions of the two problems are essentially the same: for each possible solution test if it is the correct one (that’s called the brute force or the axe method ). The difference in performance is generated by the sheer discrepancy in the solution space size (solution space is the set of all possible solutions).

Another thing that we can tell is the fact that the structure of the solutions is different: in the P case the solution parts are essentially independent (only one part) but on the NP problem a solution can have a lot of interdependencies (every part counts and can transform an ugly solution into a perfect one and vice versa). The complexity of the NP problem seems to be generated by this interdependency of the parts.

Now let’s apply this to business ideas . A good business is the result of a complex set of factors: good idea,  good environment, good opportunity, good team, etc. Every one of these factors depend on a whole set of other factors (the generation of a good idea depends on the ability to understand that there is a certain problem to be solved, or certain opportunity to change the status quo, or the ability to see that a small bend in the rules can add some value, or the ability to see that someone will have a problem in a certain period of time, or any combination of those factors). The same apply for every other factor. Basically in order to create a good business you need to be able to search a local optimum in the space of all possible business idea.

The way this is done at the moment is via a process that is applied in computing from some time. There is a certain class of solution space algorithms that work towards finding a local optimum: hill climbing, simulated annealing, etc. Basically you start from a certain state (some factors) and you work towards your goal by and adapting and changing one factor that seems to take you more closely towards your goal. The “seems” word is the key here. He makes the difference in general between classes of algorithms. There is a risk with those types of algorithms. They usually tend to get  stuck in a local optimum, a state where some factors are good but some other factors are very broken. The solution is almost good but without a drastic change in some factors there is little chance of reaching a better solution.

In the entrepreneurial world the process recommended at the moment is called Customer Development (originally appeared in the Four Steps to The Epiphany book @ Steven Blank). It’s an iterative process of proposing a prototype (or theory) and testing to see you can find paying clients, if you don’t than change the prototype and test again. The prototype can be only a descriptive solution of your product / service / shtick but the idea is to test your assumptions against the possible clientele (basically optimizing your idea). Similar principles apply to all the factors of a business: test the team to make sure it can deliver what it should if not change / optimize, optimize the environment by using technology to optimize cost / maximize opportunity , optimize the opportunity by expanding/changing/focusing and different industries, etc.

But once you see a good business implementation you can tell if it is working or not instantly. (hint .. you check the bank)

So a business build is like a basic search in a very big solution space. The problem with this type of search in infinite (or extremely big) solution spaces is that you can’t really cover a lot of cases / variants and it will always be a possibility that some kid has a little different idea of what you think is the known about the current state of affair and it will change the game.

What i want to say is the following:

Finding excellent ideas and implementing them is NP – hard .

Some might get lucky and find a better local optimum than others simply because they looked a little different at the world. There is a very small chance for someone to find the final idea and we should not focus on looking for the final idea and to start building something.

Now if you are someone who enjoys seeing the world (but not via a tour bus/train/etc) go check out globetrotter.com. Hopefully you will understand what i was trying to say .

Mihai

Bookmark It