Bonjour lets computers find devices on a local net using DNS protocol, over broadcast UDP packets, which is VERY simple. Bonjour is a small, decentralized extension to an existing protocol that buys some very useful functionality.
UPnP not only tracks where devices are but has sub-protocols for using various devices. One popular device is a router, and UPnP has a protocol for telling the router to map an outside port to a port on an inside machine, which can be bad. But if you want to do it (like for BitTorrent), it is so difficult that only obsessed evil hackers have been able to figure out how. Bonjour doesn't get involved with this.
UPnP uses SOAP, which is procedure calls and returns formatted in XML over HTTP over TCP, which is VERY COMPLICATED. Just to find out where your router is (e.g., 192.168.1.1), both you and the router have to talk to a central server. UPnP is a whole family of new, centralized protocols, and the only thing people use it for so far is something security experts say is risky.
Both these diagrams leave out Carl Hewitt's Planner and MicroPlanner, which influenced Prolog, and his Actor languages like PLASMA which influenced Smalltalk, Conniver and Scheme.
Actually the Xanadu royalty system is pretty simple. I think it would have been workable. There have been various versions of Xanadu and related royalty schemes, but they all have the property of being flat, and triggered only when actual stuff is fetched.
There are a lot of web pages with links that lead eventually, indirectly out onto most of the web and back to themselves. But your browser is always just fetching certain bytes from certain servers. The royalties would be paid on the fetched bytes when they were fetched, and the rates aren't based on how you got to the page. That is, in Xanadu-like schemes, there are no royalty-splitting, indirect credit calculations or kickbacks. A linker only gets credit if you read his own commentary (e.g. the blue text).
It's not more complicated with inline quoting instead of links: like with images in web pages, quotes would be fetched (at least virtually) from various sources, but still the royalties are being paid to those sources, for only the fetched bytes, as those bytes are fetched.
On Xanadu, the storing, caching, mirroring and credit and royalty handling were all handled by a coordinated distributed network, but looking at the web you can see that letting the individually-owned servers deliver content and charge royalties is also plausible.
Here's one area where Ted left the design deliberately slightly too simple, a solution to 95% of the problem that everyone could understand and either live with or augment with their own designs.
It's like xerographic, the generic term that Xerox made up for...the class of things the Xerox copier is an instance of. Or "Scotch brand transparent tape." You have to come up with a generic description when you apply for a trademark.
In Xanadu (the software project) lore, bad things that happened, bad omens, or bad ratings on other's writings, were "porlocks," while the positive marks were "rosebuds," I think. So instead of a -3 or +3 you could receive 3 porlocks or three rosebuds.
The Xanadu design was decentralized among a network of mutual cache/mirror servers, something like Akamai or a server-level bittorrent. There was nothing impractical in the design of it (except getting the software finished!).
The server network was going to be owned by a single company. When you published, you signed an agreement with that company to let anyone quote you with the provision that you received automatic credit (backward link) & royalties.
The company itself didn't have censorship or filtering functions, it was more of a common carrier than most ISPs are now.
I don't see why the single server owner necessarily would lead to a civil liberties disaster. At least, not more than AOL, Verizon and eBay are civil liberties disasters.
It was local governments that brought us the Bell Telephone monopoly. We are just getting free of that mess. (Of course, now we're in a new mess based on the FCC's idea of what competition means (forced competition), but I prefer it.) Local lines deregulation happened much later than other phone deregulation because local governments would not change.
The idea that a local government monopoly ISP would be more responsive than a commercial one in a competitive situation, is causing both my eyebrows and my sides to hurt. Should I laugh or cry?
By the way, government control of the streets themselves is also a hindrance here. If someone *owned* the street, and was free to set rates and charge phone and cable companies for installation and rent, then the right equilibrium between redundant cable and cable sharing would develop. As it is, the current rules interfere with the needed incentives.
The idea that redundant cable is "wasteful," is like the old USSR's gloating about the "waste" of three or four redundant US car companies. Anyway, where I live, we don't seem to be overwhelmed with cable and phone alternatives yet.
The choice between government and AOL is a false one. The point is to have a choice between competing companies. Even in roads. But networks aren't like roads, they can be built in parallel. And the idea that "the government isn't out to profit off you" is pretty naive.
The pitch of the article, (as someone else pointed out) was as if certain ideas were the last bastion of all that's good, and these are under attack. Specifically, End-to-end (and its specific current incarnations, like TCP flow control), a single flat IP space, and "Universal Service."
(It's funny, because I thought everybody agreed the internet was a fantastically lucky lean-to of kludges, in desparate need of some paradigm-shifting soon. Oh, and I thought bandwidth was cheap by now.)
The article paints the existing internet paradigms as threatened by things people want to do and are doing to networks. Each of these new activities was put in its most paranoid light. (I love the use of the word "nonconformant" meaning "not-e-to-e-correct"!) But some of them (like reliable phone connections! How hard can this be to understand!?) are just sensible applications of networks that people would like. The internet is under attack by reality. I won't address all the paranoias and stressed paradigms separately.
The internet does have a kind of unified technical-political coolness. John Gilmore said it "routes around censorship". Only here, we're talking about practices that (it is feared by some) *break* the internet. In other words, internet as it is *fails* to deal with these issues. Why? Why couldn't it serve the sensible needs and defang the bad guys, the way it "routes around censorship?" (Answer: it's imperfect.)
There's a big all-or-nothing, us-or-them attitude in both the writer and (seemingly) the other attendees. Why must it be TCP/IP-style decentralization vs. some imagined centralized architecture? What about variations on decentralization? What about fine-grained market models, for instance?
What we want is a system that can provide various sorts of services (e.g. bandwidth to some, reliability to others), to a public that is totally willing and able to pay, in a way that's more resilient in the face of bugs, clogs and the various behaviors of regulators, big companies, cops, hackers and clueless sysadmins. A way that makes bad guys less relevant and paranoia less necessary. Like real markets do. Like the internet does *somewhat* already, only with some flaws and vulnerabilities fixed.
Einstein said everything should be as simple as possible, but not simpler--I don't think pride in the internet's successes should blind us to its oversimplifications and prevent us from redesigning.
The trouble is that the ideals of the people who retrench ideologically, will be "routed around" by people like Cisco and Sprint.
Slashdot Mirror
What is it with people using "inflection point" when they mean "turning point?"
The inflection point in a growth curve is where the growth goes flat and starts to slow down.
UPnP not only tracks where devices are but has sub-protocols for using various devices. One popular device is a router, and UPnP has a protocol for telling the router to map an outside port to a port on an inside machine, which can be bad. But if you want to do it (like for BitTorrent), it is so difficult that only obsessed evil hackers have been able to figure out how. Bonjour doesn't get involved with this.
UPnP uses SOAP, which is procedure calls and returns formatted in XML over HTTP over TCP, which is VERY COMPLICATED. Just to find out where your router is (e.g., 192.168.1.1), both you and the router have to talk to a central server. UPnP is a whole family of new, centralized protocols, and the only thing people use it for so far is something security experts say is risky.
For a sense of the mindsets, look at the explanation of bonjour on Apple's page
http://developer.apple.com/networking/bonjour
Then compare the description on the UPnP organization's site:
http://upnp.org/
It's not just the methods but the goals that are different. Like, er, apples and oranges.
--Steve
And, in Asimov's story "The Last Question," the Big Crunch produced the Big Bang.
Both these diagrams leave out Carl Hewitt's Planner and MicroPlanner, which influenced Prolog, and his Actor languages like PLASMA which influenced Smalltalk, Conniver and Scheme.
Actually the Xanadu royalty system is pretty simple. I think it would have been workable. There have been various versions of Xanadu and related royalty schemes, but they all have the property of being flat, and triggered only when actual stuff is fetched.
There are a lot of web pages with links that lead eventually, indirectly out onto most of the web and back to themselves. But your browser is always just fetching certain bytes from certain servers. The royalties would be paid on the fetched bytes when they were fetched, and the rates aren't based on how you got to the page. That is, in Xanadu-like schemes, there are no royalty-splitting, indirect credit calculations or kickbacks. A linker only gets credit if you read his own commentary (e.g. the blue text).
It's not more complicated with inline quoting instead of links: like with images in web pages, quotes would be fetched (at least virtually) from various sources, but still the royalties are being paid to those sources, for only the fetched bytes, as those bytes are fetched.
On Xanadu, the storing, caching, mirroring and credit and royalty handling were all handled by a coordinated distributed network, but looking at the web you can see that letting the individually-owned servers deliver content and charge royalties is also plausible.
Here's one area where Ted left the design deliberately slightly too simple, a solution to 95% of the problem that everyone could understand and either live with or augment with their own designs.
(Trying for score "boring but informative"):
It's like xerographic, the generic term that Xerox made up for...the class of things the Xerox copier is an instance of. Or "Scotch brand transparent tape." You have to come up with a generic description when you apply for a trademark.
In Xanadu (the software project) lore, bad things that happened, bad omens, or bad ratings on other's writings, were "porlocks," while the positive marks were "rosebuds," I think. So instead of a -3 or +3 you could receive 3 porlocks or three rosebuds.
The Xanadu design was decentralized among a network of mutual cache/mirror servers, something like Akamai or a server-level bittorrent. There was nothing impractical in the design of it (except getting the software finished!).
The server network was going to be owned by a single company. When you published, you signed an agreement with that company to let anyone quote you with the provision that you received automatic credit (backward link) & royalties.
The company itself didn't have censorship or filtering functions, it was more of a common carrier than most ISPs are now.
I don't see why the single server owner necessarily would lead to a civil liberties disaster. At least, not more than AOL, Verizon and eBay are civil liberties disasters.
The Treo is also GSM-only. And it is EITHER equipped for the U.S. GSM frequencies OR the rest-of-the-world GSM frequencies.
--Steve
It was local governments that brought us the Bell Telephone monopoly. We are just getting free of that mess. (Of course, now we're in a new mess based on the FCC's idea of what competition means (forced competition), but I prefer it.) Local lines deregulation happened much later than other phone deregulation because local governments would not change.
The idea that a local government monopoly ISP would be more responsive than a commercial one in a competitive situation, is causing both my eyebrows and my sides to hurt. Should I laugh or cry?
By the way, government control of the streets themselves is also a hindrance here. If someone *owned* the street, and was free to set rates and charge phone and cable companies for installation and rent, then the right equilibrium between redundant cable and cable sharing would develop. As it is, the current rules interfere with the needed incentives.
The idea that redundant cable is "wasteful," is like the old USSR's gloating about the "waste" of three or four redundant US car companies. Anyway, where I live, we don't seem to be overwhelmed with cable and phone alternatives yet.
The choice between government and AOL is a false one. The point is to have a choice between competing companies. Even in roads. But networks aren't like roads, they can be built in parallel. And the idea that "the government isn't out to profit off you" is pretty naive.
The pitch of the article, (as someone else pointed out) was as if certain ideas were the last bastion of all that's good, and these are under attack. Specifically, End-to-end (and its specific current incarnations, like TCP flow control), a single flat IP space, and "Universal Service."
(It's funny, because I thought everybody agreed the internet was a fantastically lucky lean-to of kludges, in desparate need of some paradigm-shifting soon. Oh, and I thought bandwidth was cheap by now.)
The article paints the existing internet paradigms as threatened by things people want to do and are doing to networks. Each of these new activities was put in its most paranoid light. (I love the use of the word "nonconformant" meaning "not-e-to-e-correct"!) But some of them (like reliable phone connections! How hard can this be to understand!?) are just sensible applications of networks that people would like. The internet is under attack by reality. I won't address all the paranoias and stressed paradigms separately.
The internet does have a kind of unified technical-political coolness. John Gilmore said it "routes around censorship". Only here, we're talking about practices that (it is feared by some) *break* the internet. In other words, internet as it is *fails* to deal with these issues. Why? Why couldn't it serve the sensible needs and defang the bad guys, the way it "routes around censorship?" (Answer: it's imperfect.)
There's a big all-or-nothing, us-or-them attitude in both the writer and (seemingly) the other attendees. Why must it be TCP/IP-style decentralization vs. some imagined centralized architecture? What about variations on decentralization? What about fine-grained market models, for instance?
What we want is a system that can provide various sorts of services (e.g. bandwidth to some, reliability to others), to a public that is totally willing and able to pay, in a way that's more resilient in the face of bugs, clogs and the various behaviors of regulators, big companies, cops, hackers and clueless sysadmins. A way that makes bad guys less relevant and paranoia less necessary. Like real markets do. Like the internet does *somewhat* already, only with some flaws and vulnerabilities fixed.
Einstein said everything should be as simple as possible, but not simpler--I don't think pride in the internet's successes should blind us to its oversimplifications and prevent us from redesigning.
The trouble is that the ideals of the people who retrench ideologically, will be "routed around" by people like Cisco and Sprint.