To my many fans:
This blog was initially a university course exercise "to convey the unique feeling of having other people comment on your thoughts and opinions..." as put by our lecturer.
I must say that the first few posts were mainly driven by the course assignment, as the blog's popularity over the web grew, however, it became a part of my life - just like it became a part of your life, fans. I dare say it changed my life at least to the extent it changed most of your lives. It has been quite a journey for all of us since the first "Hello world", through the frustrated "I said, hello world!" until here and we all grew and learned something about ourselves. I learned of the almost infinite world of the web and it's inhabitants which in turn learned of me.
If this sounds like closing tones to you, I must admit you know me quite well by now (not that I'm surprised). OK, I'll just say it... I have been offered a weekly column in a major NYC newspaper and am currently considering my posting future. Sorry to drop the big news like this, but there is no easy way to say it: Simple Not Simplistic may have reached its final step.
Of-course, nothing is finalized and I have my lawyers working in shifts with the tough negotiations regarding the benefits, TV interviews and t-shirts. And the newspaper business has seen better days. And I'm pretty sure my wife will veto the whole thing. But there is a chance, so this is the heads-up for you fans.
Now like Lisa Simpson, the effort of writing has left me lightheaded, so i close by saying: Google is bugging me regarding the space used-up by your comments, so please keep it simple. Not simplistic.
Friday, July 31, 2009
Monday, July 20, 2009
Tel Aviv
After the many responses I got (2 - one of which is my own test to verify the bloody thing works) for my previous posts, I decided it's about time you people got to know me a bit better.
As you read this blog religiously, you must know I'm from Israel, Tel-Aviv to be precise. At least currently. If you hear of a good 2 bedroom apartment with an external balcony and reasonable parking, please please comment. Please. Also a balcony with no apartment will be considered.
Tel-Aviv is awesome. It's young, fun, rich-cultured, liberal, and very much alive. But not all is well in the land between Bat-Yam and Herzelia. There is an ongoing process driving young people (as I consider myself still) out and rich people in. The reason is the rent rates, not to mention the cost of the apartment.
It seems that in the last 5 years (since I moved here from Beer-Sheva - another Ben fun fact) prices never stopped climbing. Just as an example - the rent rate of my apartment climbed by 150%!
As a result, the young people that make the city what it is can no longer afford it and that really sucks. When these people grow up a bit, get married (or not - very liberal town) and have kids, you have to either own an apartment (from the family or from the time the prices were sane) or be really wealthy to stay here. I'm not sure whether the same is happening in all popular cities around the world, I just know that while the world in a recession and real-estate prices are going down all over, in Israel their going up and Tel-Aviv is the front-runner.
In the last years, there were literally no new construction that are aimed for young families in Tel-Aviv, almost all new buildings are luxury apartments, which in-turn result in raising prices of the older "affordable" apartments. That also sucks.
If nothing changes, Tel-Aviv will become a boring, old and empty city. Not to mention, I may have to move away.
AS usual, this came out a little whiny, but I insist - I am not bitter!
Now you know me better.
My crapy Tel Aviv apartment
(just kidding)
As you read this blog religiously, you must know I'm from Israel, Tel-Aviv to be precise. At least currently. If you hear of a good 2 bedroom apartment with an external balcony and reasonable parking, please please comment. Please. Also a balcony with no apartment will be considered.
Tel-Aviv is awesome. It's young, fun, rich-cultured, liberal, and very much alive. But not all is well in the land between Bat-Yam and Herzelia. There is an ongoing process driving young people (as I consider myself still) out and rich people in. The reason is the rent rates, not to mention the cost of the apartment.
It seems that in the last 5 years (since I moved here from Beer-Sheva - another Ben fun fact) prices never stopped climbing. Just as an example - the rent rate of my apartment climbed by 150%!
As a result, the young people that make the city what it is can no longer afford it and that really sucks. When these people grow up a bit, get married (or not - very liberal town) and have kids, you have to either own an apartment (from the family or from the time the prices were sane) or be really wealthy to stay here. I'm not sure whether the same is happening in all popular cities around the world, I just know that while the world in a recession and real-estate prices are going down all over, in Israel their going up and Tel-Aviv is the front-runner.
In the last years, there were literally no new construction that are aimed for young families in Tel-Aviv, almost all new buildings are luxury apartments, which in-turn result in raising prices of the older "affordable" apartments. That also sucks.
If nothing changes, Tel-Aviv will become a boring, old and empty city. Not to mention, I may have to move away.
AS usual, this came out a little whiny, but I insist - I am not bitter!
Now you know me better.
My crapy Tel Aviv apartment(just kidding)
Friday, July 10, 2009
Book review - Linked: the new science of networks \ Albert-László Barabási
I first stumbled upon this book in UPA Israel's 2008 usability day conference held in the Open University campus in Ra'anana, where Prof. Sheizaf Rafaeli gave a lecture about networks and recommended this and a couple of other books as an introduction to "the new science of networks".
As someone who was utterly unaware of this exiting field, indeed this book is a good introduction. I dare say that this book at least has the potential to change the way I think about the world around me and especially about the more complex aspects. In this sense the book's declared goal is accomplished: it made me think about networks. The book discusses issues like how are networks created and how they evolve giving the reader a network-oriented view of nature, society and economy. Acting as a new framework to understand issue like democracy, the internet and its vulnerability and the spreading of viruses or new ideas, the new science of networks is a set of notions worth spreading to the masses.
Albert-László Barabási is a Hungarian physicist (proudly mentioning his home town and other Hungarian network pioneers) who together with his research team, conducted many studies in the field of networks in the University Of Notre-Dame, Indiana, USA.
Barabási claims that the 20th century reductionism approach in scientific research has almost reached the end of the road when it comes to trying to understand the world by setting it apart to its components and underlying mechanisms. Now that it's apart it's so complex that no one has any idea how to reassemble it back together to understand the big picture. The network approach is suggested to bridge the gap.
The book chapters are set as a "chain of links", gradually and coherently building the reader's knowledge, starting from the history of the graph theory and ending at the most "popular" networks: the internet, the World Wide Web, proteins in the human cell and the economy. The following is the main idea:
What are networks?
A network is a set of nodes connected by a set of links. Typically all nodes in a single network are interconnected either directly (direct link) or indirectly (through other nodes and links).
Graph Theory and random graphs
At the beginning of the book, a historical overview of graphs and the work of the Swiss 18th century mathematician Euler and the 20th century Hungarian mathematicians Erdos and Renyi is given. While Euler is associated with the foundation of a new branch in Mathematics – the Graph Theory, Erdos and Renyi took it one step ahead to form the Random Graph Theory which has dominated the field for over 50 years. In random graphs each node is connected to a fixed number of other nodes in a random fashion. For example, in a random graph where each node has two links, there will be two neighbors randomly connected to each node. However this kind of static network is rarely found in nature.
Six degrees of separation
In the 1960's and 70's studies conducted by researchers including Milgram, Granoveter and Watts demonstrated that in many highly connected real-world networks, clustering effects can be found and quantified in a way that cannot be explained by the Random Graph Theory. One of the most interesting studies is Stanley Milgram's small world experiment examining the average path length for social networks of people in the United States. In the experience letters were sent to a number of randomly selected persons from Omaha, Nebraska and Wichita, Kansas each having a name of a contact person from Boston, Massachusetts. In the letter, the receivers were asked to add their name to the letter and send it to the contact person if they knew him or her or to someone else who may know that contact person. This resulted in a chain of correspondence where the average number of leaps from the initial receiver to the contact person (path length) was 5.5. The notion known as the "six degrees of separation" has since been widely populated, in essence saying that with only 6 leaps any two persons on Earth can be linked through the network of acquaintances. Similar effects have been also found in neural paths, the World Wide Web, the physical Internet, ownership of companies, food webs and cell biology. These studies demonstrated that the topology of these kinds of huge networks cannot be described by a static random network.
Hubs and connectors
Studies conducted by Barabási's research group and others found that many these real-world networks self-organize into hub-based networks, where the hubs are a small number of nodes containing many links while the other nodes in the network have only a few links. An intuitive example is the network of airports and flight destinations. While there are many small-town airports everywhere, they usually offer connections to a small amount of destinations, while there are also very busy airports that are connected with flight to a huge number of destinations (e.g. Hartsfield Jackson in Atlanta, Frankfurt International Airport, Heathrow in London, etc.). Studies indicate that also the World Wide Web is a hub-based network where a few web-pages are pointed to by many links, while most web-pages have just a few. Hub-based networks are highly inter-connected through the hubs. In these networks, therefore, the "small world effect" is inherent. The path between any two nodes in a network is proportional to the network's connectivity. If the network has an average of k links per node and N nodes, the average path length will be logk(N).
Power laws
The distribution of many characteristics in nature is a Gaussian distribution, perhaps more commonly known as "bell curves". This usually describes random events (e.g. a person's height) that are more common around the mean value and decrease exponentially towards the edges. It might be expected that also the number of links in networks would be distributed in this manner, but as Barabási found, this is not the case. Regarding the number of links in real-world networks, Barabási found that the distribution behaves according to "the 80/20 rule", first recognized by Pareto, an Italian economist who observed at the beginning of the 20th century that 80% of the land in Italy is owned by 20% of the population. In mathematics, the "80/20 rule" is described by Power Law distribution. Unlike the "bell curve", the power law distribution has no peak but is a curve in which a few large events coexist with many small events. In 1999, Barabási's team found that these power law curves are consistently found in numerous large networks. In these networks the number of links in a small fraction of the nodes is "off the scale" of the large majority of nodes; such networks are called "scale-free networks".
The formation of scale free networks
There are two conditions that must exist to allow the formation of scale-free networks:
1. Network growth – The network is constantly growing by adding new nodes
2. Preferential attachment – the network's new nodes are more likely to be attached to existing nodes with more links
This brings forth another typical behavior of scale-free networks – the rich get richer. Since nodes with many links are more likely to be attached to new nodes, they get richer with links. This poses a clear advantage to age – the older the node is, the more time it had to gain links from newer nodes. For example, in the scale-free network of journal article citations, early article publication on a certain subject is more likely to be cited, once it is, it is even more likely to be cited again in new articles and eventually can become a hub in the citation network.
It would seem like new nodes have a merely small change to become hubs, the explosion of Google at the turn of the century forced the modification of the suggested model. Barabási introduces a third principle to the formation of scale free networks:
3. Fitness model – new nodes that present competitive advantages are more likely to be linked.
Examining the fitness model data, Ginestra Bianconi, one of Barabási's graduate students noticed that the calculations used were very similar to those found in the formation of a Bose-Einstein condensate. The mathematical description of the behavior of "Bose gases" turned out to be identical to those in the network fitness model. According to Barabási, this means that in certain circumstances, particularly fit nodes in a network can not only get richer, but "winner take all" by destroying the hubs to form a "star network" where there is only one hub to which all other nodes are connected, this being their only link. The one example of this kind of network given by Barabási is the network of PC operating systems where the OS's are the nodes and the PCs are links – the obvious "winner take all" here is Microsoft with the Windows OS.
Robustness and vulnerability of scale-free networks
The scale free networks that exist in nature present a remarkable ability to survive under a vast range of conditions due to the high inter-connectivity between the nodes and the scale-free topology. The power-law distribution implies that the majority of nodes have only a few links; this means that for random failures nodes with small connectivity will be selected (and destroyed) with much higher probability. The removal of these "small" nodes does not alter the path structure of the remaining nodes, and thus has no impact on the overall network topology.
Of course this also presents the Achilles Heel of the scale-free networks – attacks on hubs. The 1996 Western power blackouts (affecting 11 US states and 2 Canadian provinces) is an example of the results of "cascading failures" in the power grid network. These cascading failures have also been demonstrated in other networks like the East Asian economy in the 1997 crisis, in ecological networks, etc. Duncan Watts's investigations on cascading failures found that such cascades do not occur instantaneously; failures may go unnoticed for a long time before starting a landslide.
The spread of viruses and fads are examples of diffusion in a network, with a calculable spreading rate. When this rate is larger than an epidemic threshold, the spread continues. It has been found that in scale free models the epidemic threshold disappears, explaining the persistence of some computer and biological viruses that should have disappeared according to Random Graph Theory. For example, the explosion of the AIDS epidemic, a disease that is considered hard to spread, is explained by its originating in a "hub in the social-sexual network" – an extremely sexually active flight attendant known as "patient zero".
The World Wide Web
The WWW is a directed network, meaning the links point one way. It has been found that all directed scale free networks have similar and analytically predictable topology consisting of four equal size "continents":
1. A strongly connected core - it gives a home to all indexed websites and is easy navigable
2. IN continent – easily navigating to the core, but with no paths back
3. OUT continent – easily reached from the core but with no paths back
4. Tendrils and islands – interlinked groups that have no links to or from the core. These are the web sites most search engines will never find (a quarter of the WWW).
Map of the World Wide Web
What do I think?
For me, as a novice to the field of networks this book was very interesting to read, opening my mind to see complex systems differently. In many cases these systems are so complex to that it's tempting to regard their behavior and events as random. Network thinking gave me a new mechanism to consider that can explain these complex systems.
This book is written in a very constructive and coherent manner making it understandable to any adult reader wishing to get an overview of a field that is likely to gain popularity in the future.
Critics of this book may say that the presentation of studies is too informal and can sometimes be simplistic. While this can bother a network physicist reading this book, I must say that I wasn't bothered by this at all. Another disadvantage is that it seems that the author sometimes goes out of his way not to write something even just a bit complicated and trying to write in a prose style.
To me it would be interesting to use network thinking in the context of innovation and product design. Networks of technology? Of requirements perhaps? Certainly worth thinking about it.
Since this was book released in 2002, the content may be not up-do-date. I would recommend future releases to have to update the content, because reading about the internet and the WWW in a 2002 perspective can be a bit anachronistic and naturally will be more so in the future, especially in the context of networks that evolve as fast as the WWW. Not doing so may result in this book losing its relevance in a few years.
Are you still here?
In conclusion, I would very warmly recommend this book to anybody who is interested in networks. Moreover, since the internet and the world wide web has an increasingly dominating role in our lives I would recommend this book to everybody.
Besides giving an overview of the new science of networks and its growing number of applications, I think this book can change the way you think about the world around you. Go get it!
Sources
http://www.amazon.com/review/R3FAO51MVVVGX
http://www.amazon.com/review/R2G7RCBDHVG314
http://www.dougsimpson.com/blog/archives/000075.html
http://www.bearcave.com/bookrev/linked/
Albert-László Barabási The Architecture of Complexity: From the WWW to network biology 2004 Princeton [PPT]
As someone who was utterly unaware of this exiting field, indeed this book is a good introduction. I dare say that this book at least has the potential to change the way I think about the world around me and especially about the more complex aspects. In this sense the book's declared goal is accomplished: it made me think about networks. The book discusses issues like how are networks created and how they evolve giving the reader a network-oriented view of nature, society and economy. Acting as a new framework to understand issue like democracy, the internet and its vulnerability and the spreading of viruses or new ideas, the new science of networks is a set of notions worth spreading to the masses.
Albert-László Barabási is a Hungarian physicist (proudly mentioning his home town and other Hungarian network pioneers) who together with his research team, conducted many studies in the field of networks in the University Of Notre-Dame, Indiana, USA.
Barabási claims that the 20th century reductionism approach in scientific research has almost reached the end of the road when it comes to trying to understand the world by setting it apart to its components and underlying mechanisms. Now that it's apart it's so complex that no one has any idea how to reassemble it back together to understand the big picture. The network approach is suggested to bridge the gap.
The book chapters are set as a "chain of links", gradually and coherently building the reader's knowledge, starting from the history of the graph theory and ending at the most "popular" networks: the internet, the World Wide Web, proteins in the human cell and the economy. The following is the main idea:
What are networks?
A network is a set of nodes connected by a set of links. Typically all nodes in a single network are interconnected either directly (direct link) or indirectly (through other nodes and links).
Graph Theory and random graphs
At the beginning of the book, a historical overview of graphs and the work of the Swiss 18th century mathematician Euler and the 20th century Hungarian mathematicians Erdos and Renyi is given. While Euler is associated with the foundation of a new branch in Mathematics – the Graph Theory, Erdos and Renyi took it one step ahead to form the Random Graph Theory which has dominated the field for over 50 years. In random graphs each node is connected to a fixed number of other nodes in a random fashion. For example, in a random graph where each node has two links, there will be two neighbors randomly connected to each node. However this kind of static network is rarely found in nature.
Six degrees of separation
In the 1960's and 70's studies conducted by researchers including Milgram, Granoveter and Watts demonstrated that in many highly connected real-world networks, clustering effects can be found and quantified in a way that cannot be explained by the Random Graph Theory. One of the most interesting studies is Stanley Milgram's small world experiment examining the average path length for social networks of people in the United States. In the experience letters were sent to a number of randomly selected persons from Omaha, Nebraska and Wichita, Kansas each having a name of a contact person from Boston, Massachusetts. In the letter, the receivers were asked to add their name to the letter and send it to the contact person if they knew him or her or to someone else who may know that contact person. This resulted in a chain of correspondence where the average number of leaps from the initial receiver to the contact person (path length) was 5.5. The notion known as the "six degrees of separation" has since been widely populated, in essence saying that with only 6 leaps any two persons on Earth can be linked through the network of acquaintances. Similar effects have been also found in neural paths, the World Wide Web, the physical Internet, ownership of companies, food webs and cell biology. These studies demonstrated that the topology of these kinds of huge networks cannot be described by a static random network.
Hubs and connectors
Studies conducted by Barabási's research group and others found that many these real-world networks self-organize into hub-based networks, where the hubs are a small number of nodes containing many links while the other nodes in the network have only a few links. An intuitive example is the network of airports and flight destinations. While there are many small-town airports everywhere, they usually offer connections to a small amount of destinations, while there are also very busy airports that are connected with flight to a huge number of destinations (e.g. Hartsfield Jackson in Atlanta, Frankfurt International Airport, Heathrow in London, etc.). Studies indicate that also the World Wide Web is a hub-based network where a few web-pages are pointed to by many links, while most web-pages have just a few. Hub-based networks are highly inter-connected through the hubs. In these networks, therefore, the "small world effect" is inherent. The path between any two nodes in a network is proportional to the network's connectivity. If the network has an average of k links per node and N nodes, the average path length will be logk(N).
Power laws
The distribution of many characteristics in nature is a Gaussian distribution, perhaps more commonly known as "bell curves". This usually describes random events (e.g. a person's height) that are more common around the mean value and decrease exponentially towards the edges. It might be expected that also the number of links in networks would be distributed in this manner, but as Barabási found, this is not the case. Regarding the number of links in real-world networks, Barabási found that the distribution behaves according to "the 80/20 rule", first recognized by Pareto, an Italian economist who observed at the beginning of the 20th century that 80% of the land in Italy is owned by 20% of the population. In mathematics, the "80/20 rule" is described by Power Law distribution. Unlike the "bell curve", the power law distribution has no peak but is a curve in which a few large events coexist with many small events. In 1999, Barabási's team found that these power law curves are consistently found in numerous large networks. In these networks the number of links in a small fraction of the nodes is "off the scale" of the large majority of nodes; such networks are called "scale-free networks".
The formation of scale free networks
There are two conditions that must exist to allow the formation of scale-free networks:
1. Network growth – The network is constantly growing by adding new nodes
2. Preferential attachment – the network's new nodes are more likely to be attached to existing nodes with more links
This brings forth another typical behavior of scale-free networks – the rich get richer. Since nodes with many links are more likely to be attached to new nodes, they get richer with links. This poses a clear advantage to age – the older the node is, the more time it had to gain links from newer nodes. For example, in the scale-free network of journal article citations, early article publication on a certain subject is more likely to be cited, once it is, it is even more likely to be cited again in new articles and eventually can become a hub in the citation network.
It would seem like new nodes have a merely small change to become hubs, the explosion of Google at the turn of the century forced the modification of the suggested model. Barabási introduces a third principle to the formation of scale free networks:
3. Fitness model – new nodes that present competitive advantages are more likely to be linked.
Examining the fitness model data, Ginestra Bianconi, one of Barabási's graduate students noticed that the calculations used were very similar to those found in the formation of a Bose-Einstein condensate. The mathematical description of the behavior of "Bose gases" turned out to be identical to those in the network fitness model. According to Barabási, this means that in certain circumstances, particularly fit nodes in a network can not only get richer, but "winner take all" by destroying the hubs to form a "star network" where there is only one hub to which all other nodes are connected, this being their only link. The one example of this kind of network given by Barabási is the network of PC operating systems where the OS's are the nodes and the PCs are links – the obvious "winner take all" here is Microsoft with the Windows OS.
Robustness and vulnerability of scale-free networks
The scale free networks that exist in nature present a remarkable ability to survive under a vast range of conditions due to the high inter-connectivity between the nodes and the scale-free topology. The power-law distribution implies that the majority of nodes have only a few links; this means that for random failures nodes with small connectivity will be selected (and destroyed) with much higher probability. The removal of these "small" nodes does not alter the path structure of the remaining nodes, and thus has no impact on the overall network topology.
Of course this also presents the Achilles Heel of the scale-free networks – attacks on hubs. The 1996 Western power blackouts (affecting 11 US states and 2 Canadian provinces) is an example of the results of "cascading failures" in the power grid network. These cascading failures have also been demonstrated in other networks like the East Asian economy in the 1997 crisis, in ecological networks, etc. Duncan Watts's investigations on cascading failures found that such cascades do not occur instantaneously; failures may go unnoticed for a long time before starting a landslide.
The spread of viruses and fads are examples of diffusion in a network, with a calculable spreading rate. When this rate is larger than an epidemic threshold, the spread continues. It has been found that in scale free models the epidemic threshold disappears, explaining the persistence of some computer and biological viruses that should have disappeared according to Random Graph Theory. For example, the explosion of the AIDS epidemic, a disease that is considered hard to spread, is explained by its originating in a "hub in the social-sexual network" – an extremely sexually active flight attendant known as "patient zero".
The World Wide Web
The WWW is a directed network, meaning the links point one way. It has been found that all directed scale free networks have similar and analytically predictable topology consisting of four equal size "continents":
1. A strongly connected core - it gives a home to all indexed websites and is easy navigable
2. IN continent – easily navigating to the core, but with no paths back
3. OUT continent – easily reached from the core but with no paths back
4. Tendrils and islands – interlinked groups that have no links to or from the core. These are the web sites most search engines will never find (a quarter of the WWW).
Map of the World Wide WebWhat do I think?
For me, as a novice to the field of networks this book was very interesting to read, opening my mind to see complex systems differently. In many cases these systems are so complex to that it's tempting to regard their behavior and events as random. Network thinking gave me a new mechanism to consider that can explain these complex systems.
This book is written in a very constructive and coherent manner making it understandable to any adult reader wishing to get an overview of a field that is likely to gain popularity in the future.
Critics of this book may say that the presentation of studies is too informal and can sometimes be simplistic. While this can bother a network physicist reading this book, I must say that I wasn't bothered by this at all. Another disadvantage is that it seems that the author sometimes goes out of his way not to write something even just a bit complicated and trying to write in a prose style.
To me it would be interesting to use network thinking in the context of innovation and product design. Networks of technology? Of requirements perhaps? Certainly worth thinking about it.
Since this was book released in 2002, the content may be not up-do-date. I would recommend future releases to have to update the content, because reading about the internet and the WWW in a 2002 perspective can be a bit anachronistic and naturally will be more so in the future, especially in the context of networks that evolve as fast as the WWW. Not doing so may result in this book losing its relevance in a few years.
Are you still here?
In conclusion, I would very warmly recommend this book to anybody who is interested in networks. Moreover, since the internet and the world wide web has an increasingly dominating role in our lives I would recommend this book to everybody.
Besides giving an overview of the new science of networks and its growing number of applications, I think this book can change the way you think about the world around you. Go get it!
Sources
http://www.amazon.com/review/R3FAO51MVVVGX
http://www.amazon.com/review/R2G7RCBDHVG314
http://www.dougsimpson.com/blog/archives/000075.html
http://www.bearcave.com/bookrev/linked/
Albert-László Barabási The Architecture of Complexity: From the WWW to network biology 2004 Princeton [PPT]
Subscribe to:
Posts (Atom)
