Could the Internet ever die? Or become obsolete? My guess is that someday the Internet Protocol will be replaced by something more efficient that people will migrate to, but it won’t ever totally go away, at least not for thousands of years. After all we still use many of the same railroads we had in the 19th century and even some Roman roads are still in use today.
The Internet evolved from something called ARPANET which became officially operational in 1975 and the Internet grew from it in the early 1980s.
I don’t know much about weaknesses or proposed improvements to IP, but as you note once things get entrenched it takes a huge effort to make changes, no matter how much improvement might be needed. The New York subway is something like 100 years old. I think the limitations are more related to transmission lines and switching hardware than IP.
For history, see: History of the Internet - Wikipedia
As for destructability…
In theory the internet is a network of networks, kinda like highways linking together smaller regional roads. You can’t destroy the whole Internet at once any more than you can destroy every road in the world at once unless you destroy the whole world simultaneously.
In practice, however, just like the highway system, the Internet is connected by big arteries. Big internet backbone companies (“Tier 1 providers” that form the Internet backbone) connect vast swaths of businesses and users to one another, and severing these connections (whether through malfuction, sabotage, or otherwise) can and often does result in thousands or millions of internet users being cut off until it’s fixed. Internationally, sometimes accidents or wars or policies can break or otherwise disable the internet lines going into certain countries and cut their entire nation off from the rest of the world. And increasingly, the Internet is defined less and less by the actual cables and more by the businesses that host data. When Amazon’s web services go down, all the businesses that use them go down as well, even if the physical cables are still there.
To deal with this sort of issue, many big organizations use redundant hosting that store copies of their data on different servers all around the world, and have fallback plans to re-route their connections if the main server(s) fail.
Nonetheless, that redundancy will usually still have bottlenecks somewhere or another. Even if a company manages to maintain internal connections to their other offices, for example, a hosting issue could cause thousands of their customers to be cut off for a while.
As for the future, the Internet’s infrastructure is built on something called the OSI model, which basically layers different languages on top of one another. The Straight Dope is sent to you via HTML which lives on top of HTTP which lives on top of TCP/IP which lives on top of several other layers, until it finally reaches blinking fibers of light in long cables. It is designed so that you can replace some layers without touching the others, such that you can replace fiber with copper cable but still keep HTML or replacing TCP/IP while maintaining some of the fiber.
Upgrades to the Internet happen continuously, as you can see by broadband speeds getting faster over time (unless you’re American or North Korean). In that sense, the Internet is more a concept than a snapshot in time, and you can no more destroy that concept than you can destroy the concept of “lots of connected roads”. Even if individual portions are destroyed, the remainder of the network can exist and become reconnected.
As an upgrade example, the current IP version is IPv4, which has a maximum of 4,294,967,296 unique IP addresses (in actuality, the real number is quite a bit less than that, because large blocks of addresses are reserved). The world ran out of IP addresses recently - the only thing keeping us from the inter-apocalypse is the fact that most devices are behind routers that do NAT, which multiplies the number of effective IP addresses. IPV6, which increases the number of IP addresses enormously (to something like 340 undecillion, or enough for every atom on Earth to have hundreds of IP addresses, and then some) is being rolled out, and is replacing IPV4 more-or-less transparently to most users.
Despite all the layers of protection, the internet (like the whole modern world) is still totally and utterly dependent on one simple thing: electricity.
In 2003, a very simple and localized malfunction in one spot on the electric grid destroyed all power —
leaving one quarter of the population of America sitting in the dark for 2 days.
What we call the Internet as a global thing that everyone uses for data interconnectivity, and the Internet as the historical technical outgrowth of ARPANet are arguably slightly different beasts.
Until we evolve psychic abilities, or invent faster than light travel, the basic needs of data interconnectivity will likely remain in much the form they have now. There are only so many ways of doing things. One place things change is when the distances get really large - where latency is so noticeable that you may want to do things in a fundamentally different manner. So if we become a more space-faring species the ground rules of communication will change.
Other than that, there are not a lot of ways you can build a global data network. There is one big difference between the Internet and some other possible data networks. The Internet (using IP protocols) is a routed network. You send a packet out with its destination address on its head, and the network words out dynamically how to get it there. The clear alternative is a switched network, where the sender and recipient need to work out a path between one another, and the data is launched down this path. Paths can be set up and torn down dynamically. A switched network is how modern voice calls are made. Switched networks have lots of neat properties that telcos cared a lot about. Things like quality of service guarantees and very low jitter and latency. But faster and faster infrastructure, and the marginalisation of voice as a payload have meant that these networks are slowly being replaced by IP based ones, where these properties really matter much less.
In principle, there might come a time technologically where switched networks become favoured, or some other routing mechanism than used by IP is preferred. As we progress to more and more tightly integrated and ubiquitous networking across the planet this may come to pass.
There are significant vulnerabilities in the Internet that could be used to bring it down. However the somewhat anarchic nature of it provides some level of resilience, making it hard to wipe out as a whole. Attacks on the name servers is a commonly mentioned possibility. Routing information has been subject to attack as well.
Another question that isn’t really understood is simple scalability. If every object on the planet is connected will the protocols scale well enough? Possibly not. But that will require evolution of the protocols, not the actual replacement of the Internet as such.
Finally, politics and business interests may reach a point where the Internet is not welcome in its current form. Political control of communications, and commercial interests in making money out of communications, may cause splits. There are already many private internets criss crossing the planet. Ones that have no connection at all to the Internet. Highly secure, these networks might use the same fibre infrastructure, but otherwise have nothing in common with the Internet. Whilst current attempts to control the Internet under some regimes are somewhat futile, they are not without impact. Net Neutrality could be argued as being a key attribute of the current Internet, and the loss of it could be argued to significantly change the nature of the Internet. Lots of people, with many different agendas would like to end it.
Wrong. The OSI Model is the philosophy that lost to the Internet; it’s embodied by a different group of protocols which practically nobody uses anymore, and it is too complex to bother with if you want to understand the Internet. The Internet Model has four layers, the OSI Model has seven.
As was already said, the internet is just a bunch of interconnected computers, and you can’t destroy all of the interconnections any more easily than you can destroy all of the roads in the world (good analogy there, Reply).
The internet doesn’t have road signs though. One vulnerability is that we like typing stuff in ways that people can actually understand, so we use URLs like boards.straightdope.com. This type of text is pretty much meaningless to a computer. It needs to be able to translate that into numbers. The way it does that is through DNS (Domain Name System). Basically, if your computer doesn’t know what boards.straightdope.com is, it sends a request to your local DNS server. If that DNS server doesn’t know, it forwards the request up the chain of DNS servers. The DNS servers form a hierarchy, and there are only 13 “root” DNS servers for the entire internet. These 13 servers contain the master list of how to translate names into numbers.
If you can somehow take down all 13 root servers, you can make it very difficult for computers to translate names into numbers. All of the network connections are still there between computers but you can’t find where anything is, except for whatever DNS entries are still cached by your ISP’s local DNS server. It’s like the roads are all still in place, but someone removed all of the street signs and road maps. If you know where to go you can still get there, but if you don’t know where to go then no one can tell you the way.
Taking down all 13 servers isn’t as easy as it used to be. As the internet grew, the 13 original machines quickly got overloaded. So now instead of 13 real servers, there are 13 virtual servers which are spread out over several hundred actual machines placed all around the world.
If you could somehow take control of all 13 root servers though, you could effectively make the entire internet mostly grind to a screeching halt. No one has managed to do that so far, but in 2002 hackers did manage to take down all except 4 of the 13 root servers using denial of service type attacks. In 2012, Anonymous threatened to take down the internet in protest of SOPA. They managed to muck up access to many financial institutions, but they came up pretty far short of their goal of taking down the entire internet.
Some of us were on the net before DNS. You could certainly make it unusable for most people.
Aside from permanently killing all the connections, I don’t know what the Internet dying even means. Changing protocols won’t do it. I suspect there are very few of the same machines on now as in 1985 - I’m sure all the Vaxen that used to route traffic have gone to their rewards. Blocking it at a few borders is not going to kill it, censoring is not going to kill it. With the internet of things it is poised to really explode.
Thank you for the correction. I was wrong, and I’m glad to learn that it’s no longer so needlessly complex.
As for the relevance of this to the OP, the benefit of layering still stands: It’s not TCP/IP or HTTP or fiber or any one specific thing that constitutes the Internet, but rather the interconnections of separate parts into one mostly unified whole. You can swap, upgrade, or even eliminate certain layers/portions and the Internet will still remain the same basic idea, a network of networks that connects computers together through a whole lot of individual linkages. The details of that implementation changes over time, but the idea stays: “Let’s connect all this shit together!”.
However, even if the Internet stays, it won’t be very useful to most people without the services that live on top of it: Facebook, Wikipedia, Google, Netflix, Amazon, etc. Those companies are probably more meaningful to most folks than the Internet itself. Without them, the Internet is just another type of phone service where you can reach specific people/computers if you know how to get to them, but without aggregation and directory services, they’re just billions of unsorted IP addresses.
No problem. I just get annoyed when people who otherwise know things bring up that dinosaur for no good reason.
There was, in fact, a political conflict between the OSI people and the TCP/IP people, for a while it looked like OSI would win because all of the important organizations seemed to be behind it, but the TCP/IP camp got working systems out the door faster while the OSI standards bodies were still working. TCP/IP is a triumph of practical engineering, and it seems revisionist and, frankly, deeply unfair for the OSI partisans to come back and say that TCP/IP is best understood as being an implementation of the OSI Model. No, sorry, the OSI Model had its chance and complexity lost to practicality.
Besides, the OSI Model makes distinctions without a difference. Who in the Hell knows or cares about the distinction between Application and Presentation layers, and why would they be relevant to a networking programmer?
Right. As I said in a previous thread, the Internet isn’t so much a specific technology as a specific activity: Linking disparate networks with different underlying technology together using some common protocol suite. Precisely which common protocol suite is vastly less important; you can change it and still have an Internet, and in fact we have changed it at least once in the history of the Internet. (NCP is either early Internet or pre-Internet, depending on where you draw the line, but the ongoing switch from IPv4 to IPv6 is definitely within the lifetime of the Internet.)
Heh. Remember when everyone was concerned that all the good domain names would be gone soon, and we wouldn’t be able to find anything because most organizations would be forced to use unintuitive ones? Ah, nostalgia. Anyway, the specific link aggregation companies don’t matter nearly as much as the fact there are such websites; Slashdot gave way to Digg, which gave way to Reddit. Similarly with the social media end of things: Friendster was replaced by MySpace, and MySpace was replaced by Facebook.
Forget all that history stuff. Pope Francis says the internet is a gift from God!
And what God giveth, Comcast taketh.
A nitpick, but:
According to the link, the blackout affected 10 million Canadians and 45 million Americans. How do you get a quarter of the population of America?
The OSI model is still very much alive as a concept. It was taught to me in college and still shows up every time I read anything network related. But it was only a few years ago that I realized “Hey! This isn’t just a theoretical framework for placing TCP/IP and Ethernet in their place, this protocol actually existed and competed with TCP for most of the 80s!”
The reason it works so well theoretically is that it’s so expansive – you can visualize or differentiate all possible network layers, even if you merge a few or don’t put them all into practice. Of course, that’s exactly what makes it less than ideal as a practical implementation.
The internet in its modern form i.e. the World Wide Web was invented 30 April 1993.
But it wasn’t until 1995 that HTML 2.0 was completed, which is when sites like eBay claim to have started.
The Internet is a network. The Web is a system of content that is transmitted using the Internet; it is not a form of the Internet. Email, ftp, sockets, etc., are also protocols for transmission of content over the Internet.
The Web has about as much to do with the Internet as Lisa Frank stationary has to do with the US Postal Service.
I don’t know if it’s the Internet’s birthday, or it’s moment of conception, but I’d nominate 10:30 PM PST on October 29, 1969, when the first message was sent from one computer to another, via ARPANET from UCLA to the Stanford Research Institute.
I find it somehow strangely comforting that the first ever network disconnect occurred just moments later; the message was supposed to be “LOGIN”, but the system crashed after transmitting “LO”