O'Reilly Media Asks: Is It Time To Build A New Internet? (oreilly.com)

An anonymous reader shares an article from O'Reilly Media's VP of content strategy: It's high time to build the internet that we wanted all along: a network designed to respect privacy, a network designed to be secure, and a network designed to impose reasonable controls on behavior. And a network with few barriers to entry -- in particular, the certainty of ISP extortion as new services pay to get into the "fast lane." Is it time to start over from scratch, with new protocols that were designed with security, privacy, and maybe even accountability in mind? Is it time to pull the plug on the abusive old internet, with its entrenched monopolistic carriers, its pervasive advertising, and its spam? Could we start over again?

That would be painful, but not impossible... In his deliciously weird novel Someone Comes To Town, Someone Leaves Town, Cory Doctorow writes about an alternative network built from open WiFi access points. It sounds similar to Google's Project Fi, but built and maintained by a hacker underground. Could Doctorow's vision be our future backboneless backbone? A network of completely distributed municipal networks, with long haul segments over some public network, but with low-level protocols designed for security? We'd have to invent some new technology to build that new network, but that's already started.
The article cites the increasing popularity of peer-to-peer functionality everywhere from Bitcoin and Blockchain to the Beaker browser, the Federated Wiki, and even proposals for new file-sharing protocols like IPFS and Upspin. "Can we build a network that can't be monopolized by monopolists? Yes, we can..."

"It's time to build the network we want, and not just curse the network we have."

Comments

With..,

[Anonymous Coward]

With blackjack and hookers!

Re:With..,

[freeze128]

...and IPV6!

Confusing wording/philosophy?

[Anonymous Coward]

"a network designed to respect privacy, a network designed to be secure, and a network designed to impose reasonable controls on behavior."

Privacy, secure and... "controls on behavior"?

"designed with security, privacy, and maybe even accountability in mind?"

Again, it speaks of security, privacy and... accountability?

I'm not arguing against this as I don't understand what is meant. I simply want to understand how privacy can work together with that last thing they keep bringing up.

Re:Confusing wording/philosophy?

[hord]

Privacy doesn't mean anonymity. With encrypted protocols it's possible to share pieces of data or perform collective actions without revealing personal information. There is still a worry of data accumulation (logging) but ideally you can identify bad actors and remove them from the system with minimal damage rather than the wild west of identity we have today.

Re: Confusing wording/philosophy?

[Anonymous Coward]

Can't speak for that guy but here is one example - let's say we want to build in defense against DDoS. It seems possible to have some network rules about respecting the destination and refusing to forward packets along a route where the destination has replied saying "stop sending me packets so fast" , the routers in between don't need to know anyone's identity in order to slow down that stream, and if everyone did this then eventually attacker can only get packets across one hop from each zombie , stopping the DDOS without breaching privacy.

Yes, service to that destination will be degraded for real folks who own the zombie machined , but that's either irrelevant (they're not accessing that site anyway) or even good (they can be notified they were asked to slow down, and coupled with fact they didnt try to access it should tip them to fact they are owned and maybe they can recover their PC).

Re: Confusing wording/philosophy?

[Anonymous Coward]

It's not all or nothing.

I have a family , I have friends, I have coworkers... If I can speak to them privately it means nobody knows what we are saying. It doesn't ALWAYS mean that nobody even knows we are talking. It's like taking someone to the side at a party - everyone can see you talking but doesn't know what you are saying. That's a limited form of privacy and it is adequate for many personal communications.

The absolute privacy you're talking about is either anonymity (someone can see you talking but doesn't know who you are) or secrecy (nobody even knows you are communicating).

That last one is what worries the NSA because it's what enemies of the state need.

Everyone else , including paranoid business folks workong on confidential projects , usually is fine with regular privacy or anonymity. Maybe some aren't but the point is that there ARE shades of this that are acceptable to many without saying it's ok for state to spy on everyone.

Re:Confusing wording/philosophy?

[skids]

Make your own wires and leave all the spies and rent seekers and thought police behind.

...and when you try to control the flood of criminal sploit traffic making those wires useless, you become the spy. ...and when you cannot afford to keep that wire working, you become the rent seeker. ...and when you decide you don't want your pipe used for something you find morally unconscionable, you become the thought police.

All I can say is, if the idea of a "new Internet" gets tried yet fucking again, there are plenty of technologies already available to implement it. Some of them are really well designed. Ignore them, because whoever implements it will select the worst of the bunch and/or roll their own amateur crap.

Re: Confusing wording/philosophy?

[skids]

As a term of art "privacy" was getting to be too much of a polyseme, so it was downgraded to a "reason for rather than a kind of security" in RFC4949.

"Anonymity" as a term of art does not exclude an unmasking ability... the loosest form of the word may be used to describe a system that only protects association of an alias with an identity by uninvolved third parties (termed "identity protection" in some protocols), and the involved third parties are allowed to include, for example, a court that may ask for an unmasking. Just saying "anonymous" is rarely going to be specific enough... an actual explanation of the parameters is needed.

So for example, if you communicate with a website "anonymously" but the website can tell you are the same person that communicated with them yesterday, that is technically "anonymous." You cannot have any meaningful form of authentication if you are using a definition of "anonymous" that prevents communicating parties from knowing they are talking to who they intend to be talking to. About all you can do in that case is provide completely public services.

"Accountability" is an essential component of a lot of services we take for granted, especially "non-repudiation" which is essential for securing business and legal transactions. Accountability involves agreeing to some rules of behavior, which are specific to the service in question.

TFA is pretty meaningless to throw such terms out there without defining the terms and parameters, and shouting about what they mean is meaningless as a result.

Re: Confusing wording/philosophy?

[skids]

Meant to add before my itchy submit finger, some of those do require a sender identity. If you have a DoS stream incoming that is not DDoS, (or a DDoS using your network as a reflector), you need to have an identifiable source to tell the ISP to blackhole. Lacking this ability, anyone with a fatter pipe than yours can prevent other people on the network from reaching your service. This is one example where a network identity is required to maintain network sanity.

Your proposal to "stop forwarding along routes" from which a backpressure message has been received would either require backbone routers to be magically connection-aware without a source identity (ATM could do so, but IP core routers mostly are not up to this task, and ATM is AFAIK still well behind IP in scaling and not getting much investment), or some sort of mechanism by which routers closer to the victim stop blocking traffic sooner than ones closer to the attacker, which would require additional state, and would be pretty slow to converge and probably subject to relapses. Not impossible, but a whole lot of technical trouble just to forgo using a source address.

Re:Confusing wording/philosophy?

[AmiMoJo]

...and when you try to control the flood of criminal sploit traffic making those wires useless, you become the spy. ...and when you cannot afford to keep that wire working, you become the rent seeker. ...and when you decide you don't want your pipe used for something you find morally unconscionable, you become the thought police.

All those issues can be overcome. Encryption and anonymization makes it impossible to be the thought police, for example. Imagine operating a node on an onion routed network - you can't determine packet content, or source, or destination, or infer anything about the identity of anyone involved with it. You have a binary choice: route everything, or route nothing.

If properly designed DDOS attacks should become ineffective anyway.

As for paying for it, the backbones are not the problem with the current model, ISPs are. However, with a mesh wireless option we might finally be able to break the stranglehold that ISPs have on last mile infrastructure. That's probably the hardest part to do. Effective and proven technology exists for all the other stuff.

Re: Confusing wording/philosophy?

[hawkeyeMI]

Wireless is hard. I run a wireless ISP in a rural area. Every time I see these "independent mesh network" pipe dreams come up, all I can do is shake my head and laugh.
Good luck, it's not going to happen. A much more reasonable idea is to run an encrypted meta-network on top of the existing infrastructure. That's been tried too (freenet, Tor hidden services) and it's not easy but it's at least feasible.

Re:Confusing wording/philosophy?

[AmiMoJo]

I'm in the UK too. BT is shit. Where they don't have competition they are very slow to upgrade, and are well over a decade behind other developed nations. They got vast amounts of money to provide universal broadband, but you still find many areas where 2Mb is all they can offer.

There were some "luxury" houses built near where I work. Starting prices around a million. Only one sold, and when you use the BT speed checker it estimates that they can only get 0.5Mb.

My house is actually served by Virgin as well, but I left them years ago because of a fault that they failed to fix for over two years. They said it was high utilization in that area, but were still signing up new customers. The BBC investigated them for that recently and found their staff still guaranteeing 150Mb in areas with known faults where people were actually getting around 5Mb during the evenings.

Fibre is just a dream, we don't even have a date for when it might be available. I mean real fibre to the premises, not to the cabinet.

Reasonable to whom?

[BitterOak]

and a network designed to impose reasonable controls on behavior.

Who gets to decide what controls are "reasonable"? What kind of "behavior" is to be controlled, and how?

Re:Reasonable to whom?

[retchdog]

it's probably not worth jumping to paranoid conclusions based on a bad summary of an idea which is barely even embryonic.

despite the shitty orwellian wording, this could be something as benign as ad-hoc peering protocols which incentivize people to not ddos, without monitoring content or mandatory registration which everyone is abstractly afraid of but really has been in place for decades.

Re:Reasonable to whom?

[Anonymous Coward]

Basically IPV6 and end to end encryption. Sorry ISPs you do not get a say anymore.

Needs more than just that.

A new internet needs to be decentralized and anonymous to the point that nobody except participants can tell who had a conversation with whom. If the government decides A is bad and they know you talked to A or did a DNS lookup for A's website, then it doesn't matter if the channel was encrypted - they can beat whatever info they want out of you.

At the same time, a new internet needs to be able to absolutely prove (if and only if you want) that you are you and whoever you are talking to is who they say they are.

Maybe there are protocols out there right now that can do these things without too much hassle?

Re:Reasonable to whom?

[martinX]

That's not shitty Orwellian wording. It is not some semantic mistake. It says exactly what they mean it to say, and the implications are as bad as it sounds.

Imagine, for a moment, the results of China having a say on "a network designed to impose reasonable controls on behavior". China has about a fifth of the world's population. Why shouldn't they get a proportionally large say?

Re:Reasonable to whom?

[AmiMoJo]

That's a ridiculous interpretation, because the stated goal of the design is to be censorship resistant and private, both which which preclude what you imply China would want.

My initial thought was to allow individual nodes on the network to, for example, block DDOS attacks. You can't have a viable network without any control mechanisms, because it has to be resistant to bad actors. Bad actors can be people trying to censor or people trying to deny you service or attacking your systems.

It's definately time

[Maxo-Texas]

The current internet has almost become worthless.

Festering with ads and malware.

Tracking everything you search for and selling that data to the highest bidder.

Re:It's definately time

[TheDarkener]

Seriously. I never get ads on my ssh sessions (well, maybe if I'm logging into an Ubuntu box and see it's MOTD)

Re: It's definately time

[PoopJuggler]

Depends on how we build it. The real morons are the ones who accept the status quo and their enslavement to the corporations.

Sure we can, but will it be different?

[Todd Knarr]

Sure we can build a new Internet. Where are the long-haul links that connect cities going to come from, though? Let alone the intercontinental links. Or local distribution when you want aggregate bandwidth greater than WiFi provides? The logistical problems with those things are what the current control issues stem from.

And do we really need a new Internet? IPv6 itself seems pretty sane, and it's possible to build new protocols on top of it (in fact if you look for a file named "protocols" (even Windows machines have it) you'll find tons of them listed). Or even just start building application protocols that require the use of IPSec encryption/authentication.

Re:Sure we can, but will it be different?

[rmdingler]

Every swinging internet user has a vote on how things work. How you browse the internet, which sites attract the volume of your time, where you shop... you're either the customer or the product, so depending on how you vote with your time & wallet, some of this shit is your fault.

Don't like Facebook or twitter? Me neither, but the voters have spoken and we're in the minority.

Re:Sure we can, but will it be different?

[AmiMoJo]

Where are the long-haul links that connect cities going to come from, though? Let alone the intercontinental links.

The current ones are fine. What we need to do is encrypt and anonymize the traffic flowing over them. In the short term the new protocols would probably piggy back on the old ones, at least until the routers get updated to handle them natively.

Or local distribution when you want aggregate bandwidth greater than WiFi provides?

That's the hardest part. Some new tech will be required. Some ISPs are already looking at municipal scale wireless. I don't think it's all that bad though, we just need to get to a point where the alternative is good enough that it forces ISPs to compete with it by being less shitty.

Unfortunately a little naive

[Anonymous Coward]

Building new infrastructure doesn't fix the trolling/abuse issues: those are governance and I'm not sure how you fix that kind of issue without adding MORE oversight instead of reducing it as the article suggests.

The other issue is that infrastructure costs big bucks.
- Think interstate haulage, inter-country haulage.
- Wifi uses shared spectrum and just won't scale to the size we need for the most common applications these days. You see this in local free nets now & even in over-subscribed public networks.
- Additionally security requires additional bandwidth and compute. The compute is inexpensive these days, but the article is suggesting lower bandwidth infrastructure: there's going to be a collision of requirements.

The last line of the article shows the depth of ignorance: 56K modems require serious telco infrastructure to terminate the calls: a 56K modem essentially can't be used by hackers unless they terminate to a telco. the best non-telco analogue speed you can expect is 33K.

Re:Unfortunately a little naive

[swb]

I can't decide if TFA is just an example of people pining for what they've lost, like old people wishing it was the 1950s again, or if its just wishful thinking.

I think the new network they wanted was the pre-web internet and even with big bucks from government and universities and a handful of private companies who essentially weren't paying attention to the resources being given away, it was kind of barely held together. Its small and cohesive user base gave it the shared values that made it congenial.

Sadly you kind of have to face the fact that its the commercialization of the Internet is whats allowed it to grow, and interconnecting more users is both a blessing and a curse, as the loss of cohesion leads to the loss in shared values.

There's no way to rebuild it from the ground up with wifi and ad-hoc technology. You might be able to build a new network on top of the old one, but I'm skeptical it can be done.

DECnet should be considered

[Tjp($)pjT]

DECnet lost out to IP. It should be reconsidered. The network was fairly easily expanded indefinitely where addresses were only bounded by specific specs for the implementation phases. The routing as to first of 1024 addresses where the next 1024 addresses under one of the first 1024, etc. Each node learned some basic weights to give its interfaces based on dynamic results of traffic passing. Could be improved over the last Phase V DECnet spec, based on modern knowledge. The architecture was not limited to address space. Any node could have 1024 sub-nodes to extend it. So no dynamic IP allocation issues. Then redo all the protocols used considering modern processors are very very fast and that human readable traffic is not required. So encrypt everything with very strong encryption. Make everything traceable to its source. If you have the keys. Lots of ways to revamp the Internet with an eye to the future. And instead of tunneling DECnet under IP, have an IP tunnel under DECnet. Or UNnet if you want to be politically correct. Done correctly I can have worldwide satellite offices and netboot a machine in Sweden from a server in Switzerland and do it in a secure encrypted manner. Can't spoof email if it is always signed and can be verified ... Can't spoof domain resolution if everything is verified and secure. Redoing the Internet? Make it secure from the start.

First World Problems

[Kohath]

I have instant access to the world's people and knowledge. But there are ads and Netflix might have to write a check to Comcast (or something equally dire).

So yeah, let's scrap it in favor of a bunch of stuff that's barely more than an idea.

The current Internet is not the problem...

[QuietLagoon]

... it is the on-ramps that are so egregiously bad. The ISPs have taken control of the Internet because they control the on-ramps, and because they can. So, what does O'Reilly propose to get around the evil ISPs?

.
Unless the on-ramp problem is solved, everything else is little more than mental masturbation.

Re:The current Internet is not the problem...

[Kohath]

This problem gets solved for most people in 2022 or so when 5G fixed wireless gives everyone in populated areas 2 or 3 more choices of ISPs.

You can bet ...

[Billly Gates]

... that if we do the MPAA, telecoms, ISPs, and media companies will be sending out their lobbyists to make sure they own 100% of before the bill is even finished. Also the NSA and CIA will want backdoors and own all the private keys.

Russia and China will make their own internet where they will be owned by their own special dirty interest groups and government agency.

Yeah great job. As crappy as what we have now at least DNS with ICAAN and much of what we have is somewhat decentralized even if the it reaks of American rule for many international readers.

The problem is not evil ISPs. It is EVIL LOBBYING by ALL governments and special interests that is the root of the problem. The USA is a bad 1st world country where it's citizens vote on evolution, abortition, in over representated districts in rural areas to help Republican votes count more and feels giving money == free speech. Go try that with a judge folks and say your honor here is free speech and hand him $100 and see how long you get before being thrown in prison!

Yet when a company does it it is their GOD GIVEN right.

Still compared to Russia, China, and India the US is still a God send but even the EU is a little dirty.

Never going to Happen!

[oldgraybeard]

The first building blocks on today's internet were put in place by very few in academia who built the equipment and setup an initial fairly simple point to point connection.

Over time more very basic protocols and capabilities and academic users were added.

And then it was let loose, to the creativity, innovation of many and the chaotic growth happened which led to today's Internet and the Web.

There is no chance the powers that be and the corporations could ever design a replacement. The complexity and demands from stake holders could never lead to a successful project.

Just my 2 cents ;)

Re:Never going to Happen!

[oldgraybeard]

Ah forgot something, "With Darpa Money" ;)

You can't have privacy & accountability

[rsilvergun]

the two are fundamentally incompatible. Privacy only matters when powerful organizations (basically government & mega corps) are abusing it. Accountability requires consequences that are enforced. Meaning no anonymity since if you're anonymous punishment can't be enforced.

Sorry O'Reilly, but there are no simple answers to the complex problems caused by global telecom network open to all commers. It's either going to be a hodge podge of solutions tailored to solve specific problems, a broken chaotic mess or locked down by the ruling class. I'm for the first option.

Re:You can't have privacy & accountability

[guruevi]

You don't have to "know" everything about a person in order to make them accountable, especially not on the Internet, it's how Bitcoin works.

There is also no need to punish anyone for what they do on the Internet, anything "bad" that can be done on the Internet is easily resolved by some form of censorship whether it's firewalling, blocking or removing the content.

The main reason why this idea won't float is because the Internet or it's protocols inherently aren't broken. Sure there is a lot of old cruft in eg. TCP/IP or FTP but modern implementations scale very well and can be done securely.

The main "problem" with the Internet sits not between Layer 1 and Layer 5, it sits with Layer 6 and 7, and most of the trouble there is owned by Microsoft and to a lesser extent Google & co (ad companies) and a bunch of shovelware (both in hardware and software) vendors. Moving to another network of any kind will not resolve it since anyone will be able to couple the two networks and it still doesn't resolve the layers causing trouble.

Re:You can't have privacy & accountability

[Opportunist]

Accountability can take many forms, and not all of them need the ability to link something to the physical you. Take any online game where you accumulate something (xp, in-game wealth, whatever). If you cheat, your account will be banned, rendering all that in-game achievements worthless. So, in a way, even though nobody can tell that superstud99 that was just banned for cheating was you, there is still accountability in effect.

Essentially, it's usually enough to threaten an investment made by a person to make him "behave".

As dumb as this sounds...

[blackhedd]

There are definitely smart people out there who are thinking about it. I know about some efforts to do it within some very limited but highly critical domains. (Think critical infrastructure, like electric power transmission and similar.)

Of course you'll never replace the incumbent commercial Internet. You won't boil the ocean either. But in the limited domains I'm talking about, the total number of really essential endpoints can be like 1e4 or even less. Compare to the Internet at whatever it is, probably nearly 1e10 by now. It's not crazy to think about replacing the networking.

Why do it? Simple: security. What aspect of security is most critical? Accountability. Until you can receive highly trustworthy remote control signals and telemetry data from, say, grid partners, you really can't say anything with high confidence about how the grid is being managed, or even about the integrity of your assets and processes.

So what's needed? Here's a few things: 1) A new networking stack. The IP suite, as astoundingly successful as it has been, is hopeless broken for industrial security. Too many holes, too much surface. 2) A new OS (!). The networking stack is too deeply interwoven with existing kernels. The new OS will be some flavor of Linux, but with the networking broken out somehow. 3) New protocols for establishing accountability. This one is pretty fuzzy at the moment, but a core requirement. 4) New apps, or at least rewritten ones. Remember, we're not talking about a billion endpoints. This will take years but it's at least conceivably possible. 5) Fighting the brutal, determined, and hyper-funded attacks of incumbent tech and automation vendors. This is the tough one, but remember the old saying: "First they laugh..."

Yes I know that critical infrastructure is shot through with automation systems built on way-back, unpatched Windows versions. That's not changing within the capital replacement cycle, which can be 40-60 years! But that doesn't mean that gateway networking devices can't be replaced in front of the automation networking.

I'm wearing my asbestos underwear, so flame away. All I can say is: keep an open mind and stay tuned.

Re:As dumb as this sounds...

[HBI]

No need for flame retardants, this kind of thing has been discussed before.

I would look for an analogue in two places - first, the onion routed darknet. Second, Fidonet back in the dark ages. Both are overlays atop an existing communication infrastructure - the first atop the internet, the second atop the PSTN. The darknet has been persecuted to high heaven because criminal enterprises are best placed there, making the whole thing appear criminal - child porn, drug sales, etc. Fido flew under the radar because the bandwidth requirements were low and no one perceived what it could do. Which was a lot, particularly in places like the Eastern Bloc.

My belief is that any attempt to create such a network today would be bandwidth constrained from the get-go and would be swarmed by law enforcement as soon as they became aware we were trying to do it, because we've evolved to a more totalized (and hence totalitarian) society than in the 80s or 90s, believe it or not. And yes, I do remember the Soviets, but by the 80s it was Stalin-lite anyway. The social controls even in the supposedly liberal West are such that free association is immediately under surveillance. Therefore, in response to your five point plan - we can do it, but with the full knowledge we'll be made to regret it later, regardless of how well we try to cover our tracks and keep our noses clean.

This is the main reason why such things aren't happening today. We've made it difficult or impossible to innovate because of whatever the excuse for surveillance is today. "Think of the children" "War on Drugs" "Terrorism"...whatever.

ObBetteridge

[Anonymous Coward]

No.

It's called IPv6

[WaffleMonster]

The Internet is almost perfect. Restoring the Internet to a network of PEERS would make it perfect. Currently most credible path forward is continued deployment of IPv6.

Remainder of authors concerns can be fully addressed by a robust implementation of RFC3514.

This isn't about a physical network

[MobyDisk]

Half of the trouble we face today with the internet doesn't require a new *physical* network. We need instead to prefer standard protocols, and stop centralizing information with big companies. That means run your email address from your own domain instead of using gmail for everything. Don't use Facebook to login to everything. Share pictures with friends over email. Put your public thoughts on your own blog instead of tweeting them. If people are interested in following you, they will use your RSS or Atom feed.

Everything these big companies are doing to mine your data and overwhelm you with useless information are inferior (but more convenient!) replacements for the standard decentralized protocols we already had.

Unfortunately, having a few monopolies control the wires is the cheapest most efficient way to build a network. Mesh networks are just not enough to span planet earth. We are only going to address the neutrality issue with appropriate regulation. As-is, the regulation stifles competition rather than promoting it.

Hell yeah*

[schleimkeim]

A new internet, without data mining and advertising.

Yes. DNS and services redo desperately needed.

[Qbertino]

I think we can all agree that most of what we use today is historically grown and more than just a little messy/haphazard. I don't know if we need to rebuild the entire internet - TCP/IP seems to be doing fine AFAICT - but a larger portion of its key services need a redo IMHO.

- DNS needs a redo, that's for sure. Whom am I paying 2 Euros a month just for an entry anyway? Namecoin uses the blockchain for naming, and that is the way to go. A state-of-the-art DNS replacement would use that and some central registration authority where you can get a batch of tokens to register/claim the domains of your choosing and be done with it once and for all.

- E-Mail. Well, being just about the oldest service ever and still in existance. It shows at every corner. Replacement desperately needed. Default built-in hard crypto signing, enveloping, all on top of a new DNS (see above). That would make spam go away in an instant and finally make E-Mail private. Add in referer prohibition, proper threading, echo-pooling and standardized non-prorpietary attachments and rendering standards and add everything else that Usenet offers that might be useful and Facebook would finally be obsolete. Facebook only exists because E-Mail is shite and FB actually is a better version of E-Mail for most people. I can't really blame them.

- Web needs a redo. True thing. The Web has outgrown HTML roughly 20 years ago. HTML / CSS today are just about unmanagable and have grown into humongous monsters and still fall short in building a neat current-day Web experience. Well-built Flash apps from 1999 still outpace and outperform websites from today - this is a problem, as it causes significant bloat in the HTML/CSS/JS department with no real performance gains. To the contrary, sites continue to bloat and ever increase in demand with no real improvement for the user. Not good.

- Offline. We need a net that takes offline into account more. This is IMHO the internets biggest downfall alltogether. Fidonet and the likes had and still have the advantage here. It would have to be something on top of TCP/IP but below the application protocols and services, AFAICT. But it's desperately needed. Especially with todays webpages clocking in at above 2MB in size on average. Insane. This allways-online thing was crazy back then and it still is today. Bandwidth is scarce and nobody needs to be online all the time. Why don't we have services that take this into account? Ok, we have (had) Usenet and E-Mail, but Web? Not really. A web replacement should take offline into account right from the get-go.

My 2 eurocents.