P2P Will Lead To Higher ISP Charges?

Lumpish Scholar writes "This Interactive Week article suggests that P2P is a wonderful thing, the direction the Internet is going ... and utterly breaks ISPs' business models, to the extent they may raise their monthly rates, or at least offer two-tier plans that will charge some users more. If true, ISPs might be seeing cost increases from two directions: more dialup ports (because users are staying on longer, so peak usage increases), and fatter pipes to their upstream or peer ISPs. On the other hand, to quote the article: "We're seeing greater decreases in the cost of the bandwidth than we are seeing increases in individual bandwidth usage." A price increase might or might not be justified, but Slashers will surely be interested if increases are coming." People have been making this arguement for a while - remember when web surfing started to become common, and people stayed on for longer, the ISPs claimed the same things.

Comments

Oh, Oh, a VC said it so it MUST be true.

[FallLine]

The hype around P2P is valid," said Larry Cheng, an associate at venture capital firm Battery Ventures. "It's changing the nature of Internet computing."

Ok guys, I conceed, all the hype around P2P is absolutely valid, not! Sheesh, it was also VCs that pitched eToys.com, ePets.com, Amazon, and many others as being worth billions. Earnings? What do we need earnings for?

Anyone remember when "push" technology was going to be the next big thing? I predict P2P is going to go the same way as push went, mostly out the window with a lot of money chasing after it. Can anyone actually tell me the benefits of P2P, over and above client-server, in actual and real world applications? Sure, we have GNUtella, Napster (sortof), and countless others, but these all basically just exist because piracy is illegal and it's P2Pishness helps skirt the law.

Now I'm not saying that P2P has absolutely no uses, but until someone comes up with a substantial and legitimate use, I see absolutely no reason to believe the hype.

Re:Oh, Oh, a VC said it so it MUST be true.

[FallLine]

Well there seems to be a lot of vagueness in terms of defining P2P and lack of clarity in examining it, especially from those who hype it. I am well aware of most applications of it and more traditional applications. However, when someone hypes P2P as being a new and earth shattering thing, they're invariably refering to Napster-like technologies. This VC certainly is.

When it comes to Napster-style applications (insofar as they are "new"), I have very specific doubts, please refer to my reply to one of the other replies to my comments, if you want to hear them. As for the vague notion of P2P that is hyped, I find they're almost always: old applications with buzzwords attached, ill-concieved ideas, or simply undefined.

Let me just briefly hit some of your bullets:

Decentralized -- no single point of failure (in true P2P, things like Napster have a central server that can fail, but Gnutella doesn't)

Well I'd argue with GNutella, every point IS a point of failure, not just a potential one. Where are the practical and feasible applications of a truely decentralized application of this nature?

Distributed -- don't have lots of people all hitting the same resource at the same time.

So what? With bandwidth costs dropping, why bother with such a hokey structure? If repositories like Tucows, mp3.com, various mirrors, and many others can survive, what is the need for it?

More information -- the contributions of each person are added to the network, not just some central server

Again, this sounds great in theory, but what does this mean in reality? First, that same information can be added to a central server, so the argument is largely an economic one. Second, do we really want "information" from people? The simple and brutal fact of the matter is that most people are not artists, are not terribly intelligent, etc, thus I question the value of a decentralized system. There is a large need for filters, to pickout things out value and to reduce the S/N ratio. Centralized techniques take to this much better...

More storage space -- probably Exabytes or Petabytes of storage space available on a large P2P network. Try getting that in one box :)

Again, costs are plummeting. What's more, even though the total amount of storage space may be greater on a widely distributed network, is the value greater (or less)? In other words, how much of the data is going to be redundant? Generally, most of it. How much of the data is going to be of sufficient quality? Very little.

In short, I have lots of questions and doubts, and so many P2P pundits are full of hot air, vague notions, and fuzzy thinking.

Re:Oh, Oh, a VC said it so it MUST be true.

[WNight]

P2P reduces the bandwidth that any one site must have available. If Napster had to serve all those MP3s it'd never have gotten big. But with P2P, each user serves up only a few songs at once, while downloading for themselves.

P2P makes sense, but only in a case where there are many things being downloaded and any given user will have some of them.

It wouldn't work for the LoTR trailer because everyone wanted to get it, and nobody had it. But if you applied it to movie trailers in general, then it would start to work.

The problem with it is that it relies on the users having content that other people want. This is okay in small communities, but if the you didn't recognize any MP3s on someone's drive, you probably wouldn't bother downloading any. The fact that you're trading a well-known group of songs makes people more willing to trust in the worth of the content.

Re:Oh, Oh, a VC said it so it MUST be true.

[Hard_Code]

We're already seeing a flavor of "p2p" in all this distributed computing stuff (Seti@home, Distributed.net, Folding@home), as well as instant messaging. Of course it wasn't called "p2p" until the p2p hype hit (usually when hype about some amazing new paradigm hits, everything on earth is recast under that paradigm...hell, Slashdot is p2p!). The legitimate uses so far have been harnessing idle computer resources, communications, and storing stuff other people don't want you to store (the very nature of which makes such a system difficult to use). But if I inhale on the P2P pipe, I can envision many different uses (many of which were promised by "agents").

Re:Oh, Oh, a VC said it so it MUST be true.

[FallLine]

P2P reduces the bandwidth that any one site must have available.

This is true in a way. However, it must be qualified. First, does it _really_ matter? Yes, bandwidth costs money, but if it is properly located and the scale is sufficiently large (which Napster et.al definetely are) it is quite cheap. The bandwidth demand also scales with consumer demand, hence there are generally ways of getting corresponding revenue to pay those costs. (i.e., banner ads, nominal fees, various marketing, promotion, etc.) Witness the success of servives like Tucows, Cdrom, and various mirroring services. They serve out gigabytes per day of relatively obscure stuff, certainly the same could be done with the more condensed demand for popular music (not just "pop", but anything that is popular enough to be signed up with a label and be listened to-- as opposed the thousands of random sharewhare programs and such). Second, Napster's databases are not all that cheap, they are an additional cost that cannot be ignored. Third, the cost of bandwidth is just one element amongst many. One of the biggest problems with Napster, and especially with more P2P oriented services, is that the searching process can be awefully cumbersome, finding the optimal server is often tough, finding quality mp3s is tough, and many other similar issues. If piracy were legitimate, centralized servers could easily solve all of these concerns.

If Napster had to serve all those MP3s it'd never have gotten big.

Then how do you explain the success of centralized techniques that serve up equally large files? If Napster is such an efficient model, why is it that this model is pretty much exclusively confined piracy? From an economic standpoint, you would fully expect the legit methods to flood the so-called P2P technology. It's not as if Napster's technology is that tough to master.

P2P makes sense, but only in a case where there are many things being downloaded and any given user will have some of them.

I really don't think it does, with current applications at least, except in the case of this kind of piracy. Though I've covered this argument above, let me add one other point. Napster is not about cheaper costs on the aggregate, the only way you _might_ say the costs are cheaper is if you count the offsetting of costs to third parties as being nothing. In other words, Napster benefits because a lot of the sharing individuals are on high bandwidth connections and do not have to pay an amount commensurate with the amount of bandwidth they consume. But this does not mean in any way that it is more efficient, just that at present time, the way things are currently configured, it sorta works. The load of the inter and intra-nets are increased relative to highly centralized methods, particularly relative to smart mirroring, hierarchical distribution, and Akamia-like technologies. Rather than downloading one or two hops off of mae-east (in the highly centralized example), you pull from some schmoes college dorm, who could be anywhere in the United States (or worse, the world). That is many more hops, less direct, and does not benefit from the economies of scale. It also depends on the "good will" (if you could call it that) of the file sharer, which I believe might well be limited to ignorance, a certain temporary sense of "community" in the piracy community, and other similar issues.

I'd also like to add that Napster, ethics and law aside, has the benefit of ignoring the issue of value creation and payment, because they're simply not responsible for having to collect or generate revenues. In other words, it's easy not to have banner ads, forms, or what have you, not just because of the different cost structure of Napster's overall bandwidth cost, but because they're not generating these songs that people find of value, nor are they supporting the generation of this value. In essence, it's an artificial, and ultimately temporary, situation. Various parties are paying for Napster's actual costs, a lot of it doesn't have to do with economy of bandwidth and/or storage at all. If Napster were put in the situation of having to actual create value, some of actual revenue creation would be a given. Once that is in place, it is hard to argue that centralized servers' tacking on of nominal bandwidth costs is much of an issue, especially when taken in light of the benefits that more traditional,client-server, technology is capable of bestowing on the user.

ISP pricing model is all wrong

[swb]

I don't see how any ISP can make any money offering xDSL for a fixed rate. I know the gamble is most lines sit idle most of the time (mine averages 58/bytes second, w/BIND, Sendmail, Apache and three PCs connected), but at the same time it only takes a a few dozen people maxing their lines out to seriously dent an OC-3, and not every ISP has that much bandwidth.

The idea of metering isn't as bad as it sounds -- one thing that keeps ISPs from rolling out serious bandwidth (1.5Mbps SDSL for everyone) is that ISPs are afraid that they won't be able to meet the demands of that much bandwidth because the marginal income outpaces the marginal cost. Charging by the packet, ISPs could easily offer as much bandwidth as you want WITHOUT having to worry about where the cash for the next OC-3 will come from.

Re:The Economics of Bandwidth

[Sabalon]

Well, under the old ISP model, you could have 500 users. You would assume, and usage statistics would back this up, that only 10% of them would be on at a given time. So, you buy enough bandwidth to handle 50-60 users. Each user gets their 56k of bandwidth and are happy.

Now, you have 500 users with 300 of them staying on and active all at once. You have to buy more bandwidth to provide the user that 56k of bandwidth that they are used to getting, otherwise you start to lose customers.

This is different from the user saying that they would like 128k of bandwidth and how much extra would that be.

Yes, a large quantity would cost more, but why should they have to pay more for their small quantity just because all these people you wooed over to your service are now using it, cutting into your profit margin.

Or you could be like my cable modem company and just not increase the bandwidth and laugh as the profit margin grows!

Re:Perr-to-peer is CHEAPER for ISPs!

[eap]

That's only if you make the assumption that the majority of peer to peer connections are between users of the same ISP. In the Real World (TM), I would guess this is usually not the case. Services like Napster generally make no distinction between a user next door to you and one half way around the world (except in ping time, and that's only if you find the file you need both locally and from someone in a remote location, which is probably still more infrequent).

Peer to peer doesn't imply any sort of geographic proximity, it is just a description of how two clients interact with one another.

Re:The role will change

[lizrd]

Which is what the original ISPs used to do. The provision of additional content/value came later.

That's kind of true. It just used to be that dail-in and receive content people were different people than the ISP. It was really a very brief period of time when the no-frills Mom-and-pop ISP really amounted to anything. Before that your ISP was your employer or it was AOL/Compuserv and now with the real ISPs going away we're left with your connection being through something like AOL/AT&T/MSN and so forth where you're paying for their packaged content plus a TCP/IP connection to the world at large.
_____________

Re:I dont get it...

[dachshund]

I dont see how me having some sort of P2P running off my DSL line is any different than me being on IRC with files offered

That's exactly what P2P is. And ISPs don't like it. They want you to take your files off of the pretty web pages, many of which (in their master-plan) will have local caching servers ala Akamai. If you look at the TOS for most cable-modem and residential DSL providers, they specifically say "don't operate a server or file sharing program." The unpredictable downstream and upstream bandwidth that P2P generates might eventually require them to spend more money.

This is really too bad for them. As long as people have these relatively powerful machines hooked up to the net, it's inevitable that they're going to use them for more than one-way downloads and web surfing. ISPs will adjust.

Re:Time for a post-ISP future?

[micromoog]

It seems to me that ISP's are becoming redundant. They have an almost 19th century business model, one based on dominating a segment of the market. Well, that kind of thinking doesn't work on the interent, and it is the reason why the biggies - AOL for example - will eventually collapse.

The model is actually a few thousand years older than that, and is just as effective today as ever. And, sorry to say, AOL is not anywhere near "collapse" . . . like it or not, they're still fantastically successful.

I forsee that when broadband comes, the telecoms companies will be in the most powerful position.

They already are. The ISPs are just (necessary) middlemen.

With broadband and more powerful computers, there is no reason why the average users computer cannot connect to the internet directly, and bypass the ISP model entirely.

Mail servers, DNS servers, address blocks, and about 400 other reasons come to mind.

ISP's, with their ridiculous dreams of providing direct media content and products, will die and be replaced by a devolved and more democratic network.

On one side, there are a lot of ISPs that offer low rates and no content. On the other side, a lot of people want the content offered by companies such as AOL.

Communicative equality in America, land of free and equal speech, could be reborn through the arrival of P2P, IMHO.

Nice sentiment, I suppose, but comletely unrealistic. Not to mention that the P2P concept has nothing to do with Internet connection service.

Missing the point.

[Edge]

People are missing the point here. So P2P will cause a surge in bandwidth usage and connect times? Look at the Internet. It is evolving. Look at bandwidth. Just as any other technology evolves and expands, so does the amount of bandwidth available. Five years ago, how many people had 512Kbps(+) pipes coming into their bedrooms. Not many. Now, cable modem and DSL connections are becoming the norm. Analog connections are certainly still the majority, but look around at the advertisements. Bandwidth is big business. People are demanding more of it. Usage has already been increasing radically. The number of people getting on the 'net is increasing at a phenomenal pace. Just like anything else, the more you sell, the cheaper it gets. The price per amount of bandwidth has plummeted over the years. The consumer costs are not going to go up. With usage and volume ever increasing, the worst that could happen is prices may not fall as radically as they have in the past. Or, you may see less of the $7 Internet access offers. BFD. People will still be willing to pay their $50 a month for their cable modems. The majority of the people out there have jobs and are not savvy enough to schedule downloads all day while they're at work. It's pretty safe to assume that your grandma will not be on Gnutella downloading the latest ICP track. But she'll still like that cable modem because it makes her marthastuart.com pop up quicker..

It'll all be okay. Really.

Um, horsepuckey.

[sulli]

Three reasons why this won't happen:

1. Cost of backbone connectivity per user keeps getting cheaper (all that OC192/768 equipment on DWDM fiber makes adding capacity way cheap)

2. It's way too expensive to measure & bill this stuff. Maybe you can measure it - but just try to link it to a biller. Major pain in the butt - and for what? A few bucks? Not worth the investment.

3. P2P is a killer app for DSL. Why do you think Speakeasy was giving away Rios? It doesn't make sense to go after a leading reason people are buying your service in the first place...

Re:Um, horsepuckey.

[burris]

1. Cost of backbone connectivity per user keeps getting cheaper (all that OC192/768 equipment on DWDM fiber makes adding capacity way cheap)

sure, bandwidth is getting cheaper but as it gets cheaper and people's connections get faster they want to transfer richer media. First it was just JPG's and GIFs. Now it's MP3s and DiVX is starting up. Next it'll be higher res, less compression. Maybe after that it'll be teledildonics or something. Oh, and don't forget there's many decades of video history that has yet to be ripped and shared.

It's just like when you buy a new hard drive. Every time I've gotten a new disk in the last 10 years I've though "damn, this drive is so huge I'll never fill it up." but I eventually do, sooner than later.

Disk space, CPU, and RAM have been increasing by the same ridiculous amounts that bandwidth but we manage to fill them up. A 700mhz G4 is many many orders of magnitude faster than the 68000 in the first Mac but a Mac OSX on a G4 feels slower than an old OS running on the old Mac. That's because software engineers do new, more complex things with the increased available power and memory. Users do new, more bandwidth intensive things given increased bandwidth.

The complementary law to Moore's law and it's memory/bandwidth equivalents are that usage increases at the same rate. The idea that after a couple more upgrades of the backbone we'll have plenty of excess bandwith for every user to do whatever they want at a low flat rate is just naive.

What's going to happen is ISP's who have structured pricing on users with intermittent, spiky bandwidth usage are going to keep their bandwidth offerings and prices the same for longer than they would have without people using P2P apps, until their profits are back. Basically they will raise their prices without actually raising prices that consumers pay, just like the way consumer product manufacturers raise prices by giving you less product in the same sized and priced container.

Burris

Re:Um, horsepuckey.

[sulli]

Emphasis mine:

The idea that after a couple more upgrades of the backbone we'll have plenty of excess bandwith for every user to do whatever they want at a low flat rate is just naive.

The key word is "low." Customers always prefer flat rates. You're right in your subsequent comment, that ISPs will charge slightly higher fees due to the increased traffic per user - but to suggest that the ISP biz will suddenly move to usage-based pricing, against all trends to the contrary, is incorrect.

People have been suggesting this for years, usually in paranoid terms. Now as before, we have nothing to worry about.

Same old, but different

[The Man]

The Internet has *always* been a peer-to-peer network, from day one. Only for a brief period of about 5 years in the mid-90s when a large number of non-wealthy new users came on board and were forced to live without bandwidth was this not the case. If you look at the design of the protocols, and the way the internet actually works, it's fairly obvious to me that it was always fundamentally designed to be peer-to-peer. Not in the hip new let's-steal-music way, but in the sense that every machine would function as both a client and a server. Remember when every machine ran telnetd? When there were no firewalls? When "PPP" meant going to the bathroom? This ain't new, folks. But I did like it better when peer-to-peer meant the Internet was a community, not just a place to steal stuff and run DoS attacks on irc servers. The model didn't really break until people started being allowed on the Internet without an Internet-supporting OS. They never joined the community, just babbled senselessly and joined AOL in droves. It's amazing how well their OS reflects their attitude and behaviour - no services provided, but a client for everything. Take, take, take, give nothing. That's peer-to-peer? Fuck this noise.

Re:Same old, but different

[burris]

uh, there have been second class non-peer folks on the 'Net for a long, long time. They used to be called "UCCP nodes." When SLIP and then PPP came out, it was a godsend; the folks who weren't at a university or a high-tech company could finally be directly connected to the net (albiet slowly and with high latency). I remember upgrading from UUCP to SLIP and it was the greatest thing ever. They gave me a whole freakin' class C for my 14.4k SLIP connection. I could play MUDs! I guess that actually made it disasterous from a getting-anything-done point of view.

Burris

Re:Same old, but different

[swordgeek]

Goddamn right.

I figured that p2p was a truly new paradigm, until I read the specs for gnutella. Low and behold, you've got a client and a server glued together, and cleverly called a 'servent.'

Client. Server. Both running on the same machine. THIS is peer-to-peer? Gee, I guess ftp/ftpd was peer-to-peer back before the term existed then! How about UUCP?

AOL and Canter&Siegel were the death of community on the internet. Anything claiming 'peer' status after that is a lie.

Re:Same old, but different

[crucini]

It's not pathetic - it's true. OS's have a philosophy deeply ingrained, and that affects the OS users. I'm not saying you can't be clueful on Windows or clueluess on Unix, but I will say that on usenet there is at least a mild correlation between luserish behavior and Windows.
Every time you interact with an OS, it is silently teaching the values of the people who created it. Take an innocuous example from Windows is the tendency to refer to 'your computer'. As in, 'this will install FOO on your computer.' Obviously, if you are logged into (excuse me, onto) the computer, you must 'own' it, right? The idea that more than one person is affected by a system administration issue simply doesn't fit into this mindset.
Windows is deeply, in it's bones, a PeeCee OS. It is completely built around the pre-network world. The windows user learns that it's normal to be a client. The windows world is permeated with the idea that servers are somehow special, expensive, rare. Windows NT charges extra for a 'server license'. Compaq and Dell make a lot of money selling huge beefy PC servers to NT-using customers. Although many people claim that NT is a resource hog, I also suspect a psychological motivation - NT admins have a need to visually differentiate their servers from PeeCees. I've notices that NT admins had a hard time accepting the fact that an old, cast-off PC became a linux 'server' that was suddenly 'important'. Not because they're against Linux, but because the natural order of things was reversed.

Time for a post-ISP future?

[Urban Existentialist]

It seems to me that ISP's are becoming redundant. They have an almost 19th century business model, one based on dominating a segment of the market. Well, that kind of thinking doesn't work on the interent, and it is the reason why the biggies - AOL for example - will eventually collapse.

I forsee that when broadband comes, the telecoms companies will be in the most powerful position. They own the fibre that travels into the consumers home. This gives them the power. ISP's are a temporary phenomenon.

With broadband and more powerful computers, there is no reason why the average users computer cannot connect to the internet directly, and bypass the ISP model entirely.

ISP's, with their ridiculous dreams of providing direct media content and products, will die and be replaced by a devolved and more democratic network.

It is better that the communications and telecoms companies have the power than ISP's, for ISP's aim to pump media and ideas into the home whereas telecoms companies are only interested in the almighty dollar.

Communicative equality in America, land of free and equal speech, could be reborn through the arrival of P2P, IMHO.

You know exactly what to do-
Your kiss, your fingers on my thigh-

Re:Time for a post-ISP future?

[Alioth]

Where I am the telcos already run the show. They are the tier one and two providers. They've bought or started the largest ISPs already, and there's no way the independents could ever compete on cost.

Where I am (Houston) things were the same way, until my current ISP appeared. This independent ISP has elbowed its way into this market, competing with the big telecos and then some. It is increasingly successful and expanding into new cities. Independent ISPs can still succeed if they have the correct business plan and have the service that people want.

Re:Time for a post-ISP future?

[gmhowell]

As long as AOL is brainless to use and install, my mother, sister, and brother will remain customers. Easy to use email and browsing. Easy to access forums and 'sites'.

More than enough for most people and still a bargain at $21 a month, or whatever they pay.

Re:Time for a post-ISP future?

[bonoboy]

Interesting thoughts. But it's a basic anarchy that will never happen. Your model seems to require everyone running their own mail server, for one. Can you see people doing that? Your model also seems to follow some ideas which IPv6 might. It would have to, because people all being handed their own address blocks is the only way you can sidestep content providers and ISPs. What do you think your plan would do to the routing tables without Ip6? It doesn't make your idea any less valid, we'd need a strictly heirarchical system. That requires some very decent top-down organisation of the address space. Geographic regulation would have to come into play. Not a big deal perhaps.

Bandwidth always costs money. Companies want to get you using their content so that you may never know the rest of the Internet exists. The Walled Garden approach is going to be a big part of the IP networks of the future, believe me. I'm not talking your Mom & Pop ISP here, I'm talking about large fibre networks with cable television services running. This is dedicated content, and has nothing to do with your anarchy model.

Where I am the telcos already run the show. They are the tier one and two providers. They've bought or started the largest ISPs already, and there's no way the independents could ever compete on cost.

Anyway, just some thoughts. The point being that it seems a nice comment with very little thought behind what an 'ISP' is these days. Anyone with an upstream connection is an ISP. not just some guys with modem banks.

Re:Time for a post-ISP future?

[Christianfreak]

Ummm nooooo. Aside from some of the other obvious reasons why this wouldn't work pointed out by other replies to your post, can you just imagine a world where all Internet is controlled by the Telco??? Seriously who in this country likes their telco? I've never heard of one that gets high marks for customer service. If we let telco's control all of the Internet connection services then you might kiss P2P good bye since no one will ever be able to connect! Its far more convientent for the average Windoze luser to deal with a smaller ISP (who knows their customer is what keeps them in business) and let them deal with the big corporate provider. Come to think of it I personally would much rather deal with ISPs than the stupid telco.

Just as an example I tried to get DSL installed a few months ago from the telco, they gave me huge promises on dates of when it would be installed, yadda yadda. That date came and went and different departments started blaming each other for the mix-up. I never got any answers and I never got DSL. Finally I simply canceled my order 2 months after the install date. that was nearly 4 since I'd made the request! Stupid telcos.

"One World, one Web, one Program" - Microsoft promotional ad

I dont get it...

[Bob McCown]

I read the article, and all I see is the quote that P2P "will break the ISP business model". I dont see how me having some sort of P2P running off my DSL line is any different than me being on IRC with files offered, or surfing, or anything else.

Re:I dont get it...

[bluntmanspam]

Most ISP's have a business model that assumes certain ratios. For example, no ISP would have one dial-up modem (or port) for every dial-up user they service. Modems and lines are expensive, and ISP's assume that dial-up users won't be online 24 hours a day. In fact, many have SA's that prohibit staying online continuously for that long.

By the same token, upstream bandwidth is usually just a fraction of the bandwidth they currently have sold to customers. This is because even dedicated high-bandwidth customers don't use all of their bandwidth all of the time. For example, our T1 customers usually only average 200kbps each at any given point in the day. That means we only have to have 1 T1 worth of upstream bandwidth for every 6 downstream T1's sold, and can still provide quality service. If every T1 or DSL customer used all of their bandwidth all of the time, most provider's upstream bandwidth would become congested, and they would have to buy more.

Generally, bandwidth for an ISP costs more than for downstream users for just this reason. If ISP's have to buy bandwidth at a closer ratio to what they are selling, their price must go up to stay profitable. This is especially true for the smaller ISP's, since they have less margin for error. Just another reason that Mom and Pop ISP's will go away.

THERE IS NO RIGHT OR WRONG, ONLY DIFFERENT PLACES TO STAND -- Death Terry Pratchett's 'Reaper Man'

"It's politics, and there is no right or wrong... but you are wrong!" -- My girlfriend debating a point.

Bandwidth screaming through the roof.

[bonoboy]

The rate that bandwidth is expanding at right now is phenomenal. The case in this part of the world is such that there are people walking around with fibre licenses with no idea what to do with them. The Southern Cross pipe opened up and that's got a combined size of 80Gig right now, I think 160Gig in a year. At least, that's the old stats.

P2P isn't anything compared to that stuff. Bring on the holographic conferencing!

The Economics of Bandwidth

[qpt]

On the one hand, it seems hardly a matter of debate that if people use more bandwidth, they are going to have to start paying more. It is a well accepted economic principle that a large quantity of a good or service will cost more than a small quantity.

On the other hand, bandwidth is an odd sort of resource. Unlike coal, iron or wheat, it is wholly manmade. While it is correct to point out that the physically devices such as routers and switches are in fact made of natural resources, this is not the limiting factor of their production.

Rather, human ability and inginuity is required for the manufacture of IP networks. To build larger networks, effort must be taken from some other task and applied to its construction.

How does this relate to ISP prices? Quite simply, the opportunity cost of network construction goes down as demand for the service rises. Thus, even though more net bandwidth will be used, the total cost will actually be less than it is now. The tradeoff, though, is that some other good or service will lose manufacturing precedence and increase in price and lose market priority. It is anyone's guess what this good or service may be.

I suppose we could always hope it is Microsoft :)

- qpt

Re:The Economics of Bandwidth

[crucini]

Interesting point, but I think the real cost of bandwidth is the cost of stringing copper/fiber from point a to point b. And within this cost, I'd guess that there's the relatively small (although actually huge) cost of telecom-style construction (poles, manholes, repeaters, etc) and the relatively large cost of dealing with political barriers. For example, how much do you think you'd have to pay to be allowed to install your own cable on all the telephone poles in a town? I'm guessing more than the entire value of the existing physical plant, because the ILEC and cable TV co. would lobby so hard against you.
So I tend to think that the cost of routers on each end of the line is a trivial cost. Maybe some RF solution will save us.

To put it eloquently: Bandwidth = Shit

[nysus]

My elderly mother pays hundreds of dollars each quarter for the privilege of being able to flush her shit into the local water treatment plant. It never makes headlines, though.

Welcome packet meters (& limits)

[crovira]

The ISPs, wether the big boyz or the mom and pop shops (who have to go through the big boyz wire, cable or fibre,) are starting to count the number of packets they send to your IP address.

They don't need to do more than count and log. Hit the limit and you'll get a message saying you're cut off. I already have.

Got a 24kbps modem, you might never hit it. Got DSL, you will probably hit it. The more you surf, the faster you'll get cut off.

Belong to an active news group, retrieve a lot of files (a typical Smalltalk image is 100MB where I work, that's a lot of MP3s,) you could find yourself cut off from the world real fast.

Unless you pay (and pay [and pay {and pay}]) for plans with different limits you'll be sucking on a dry pipe.

Not always the case

[localroger]

My telco is my ISP, specifically BellSouth FastAccess DSL.

I went with BellSouth precisely because they own the wire. Sure, Telocity may have better customer service, but who is Telocity gonna call when the line goes bad? Riiiiight.

And in all honesty, I'm pleased with the service. I ordered the self-install kit and got it running with minimal trouble. I received the kit about 10 days after placing the order, and I was online about 4 hours after receiving the kit. This included making it work with my home network even though BellSouth specifically doesn't support home networking. I just went through the normal install for Internet Connection Sharing (yeah, I'm using Win98, sue me).

Recently I had a problem with my regular phone connection -- no dial tone. (Oddly, the DSL continued to work throughout this episode.) This required BellSouth to send out a tech, and replace the line running from the pole to my house. He arrived on time, worked like a dog, kept us informed of everything he found and did, and corrected the problem. Telcos are not necessarily evil, though the procedure for putting in the service call was pretty anonymous and bureaucratic. I groaned when they said "we'll check it within 36 hours," but then they actually did.

Why a separate line for DSL?

[localroger]

My DSL shares happily with my voice fones. You do have to put the filters on all your voice equipment as some (but not all) of them will interfere with the DSL. I have not noticed any bandwidth hit when the voice fone is in use and I regularly see the 1.5Mbps down / 200Kbps up I was promised when I signed up.

EXACTLY!

[green pizza]

Very, very well put!

Way too many people believe in the Free Lunch concept. The harsh truth is free meals don't last more than a couple of days, if they're offered at all. My ISP is the local telco (a rural telephone cooperative) and is the only one in the area offering DSL service. 2.2 Mbit ADSL was first offered by them for $60/month. The price soon raised to $90/month. About two months ago they explained that to feed their DSL customers alone was costing them 2x what they were bringing in from monthly charges. Rather than further increase the monthly cost, the ISP has chosen to no longer have any guarantees of thruput as their current customer base is already saturating one OC3 and one T3 circuit. Their monthly cost for those circuits to Sprintlink and Cable & Wireless, as well as their equipment upkeep and tech support has been to too high, even with hundreds of customers.

The simple fact is, the per-customer cost to them is higher than what they can reasonably charge without sacrificing service. They're not venture capital funded, they're a local telephone cooperative and cannot afford to bleed that much money.

To answer a FAQ, yes, they do offer lower-thruput DSL services at much lower costs. 384/128 Kbit ADSL is offered for $15/month + equipment rental or purchase.

Let the fee games begin

[Kefaa]

In many areas the local monopolies have driven competition right back out the door. I get my DSL through Verizon the ONLY game in town. While they do have increased costs for the benefit of always on, or nearly always on connections, you will have to forgive me if I do not feel they are losing money on this deal. Aside from most people shutting down their connections, ISP can/do enable software shutdown/wakeup so a dead connection is truly a zero resource user.

The price "scare" is very similar to what the cable companies did a while back. First mention you have to have a large increase [Sorry the price is going up $20/month]. Let everyone get outraged at the ridiculous increases and get the utilities commission involved. Then "relent" to a meager increase of $8. Everyone thinks they win.

--The problem is not that most people are sheep, it is that most people think they are wolves.

Tough times for ISPs

[ishrat]

"ISPs might be seeing cost increases.."

These guys already seem to be in a soup. Here's an example.

High-speed ISPs nationwide have faced many difficulties providing competing DSL service with data competitive local exchange carriers (DLECs) and Baby Bells.

Excuse me.

[mindstrm]

I don't see someone who is paying for 'unlimited network access' who actually uses it 24/7 as 'abusing' their service, or being 'unfair' to others.

Just because my idea of internet use is different than yours doesn't mean I'm 'abusing' it.

@home loves doing this. They say it's 'hundreds of time faster than dialup' and 'unlimited' and 'always on' and then they freak out if you use too much bandwidth.

== P2P hype?

[FallLine]

I don't think so. I view SETI@home et. al and IMing as simply outside of the scope of the hype. The SETI@Home-type efforts are really not commercialized or something that the average consumer is interested int. And instant messaging is an old concept that has been around for ages. Either way, it's not part of the hype. The hype of focused around Napster-type technologies and/or GNUtella-type technologies, which presumably bring a lot of people _new_ things that they desire.

As for slashdot, that's simply ridiculous. Slashdot is no more P2P than any public forum. The only thing that is P2P about it is that the users, the clients, are generating content, but the same can be said for any number of database applications. It's still the same old client-server technology.

Do the math

[green pizza]

Short of getting muliple OC192's from the same provider (good luck), there is almost NO WAY to get around having to spend $300 - $500 per megabit for access to a Tier 1 backbone provider (Sprintlink, UUNet, etc).

Say your ISP is in or near a major city and they can get a T3 or OC3 for $300 per megabit (155 x $300 = $46,500 per month). 155 Mbit, perfectly distributed and not oversold, can feed 103 1.5 Mbit DSL users. However, done that way, it's costing the ISP $450 per user per month. You're not going to see that happen. So, ISPs usually oversell by 10 - 20x, which will drop the per user bandwidth cost to, say, $45 at 10x overselling.

So, *excluding* their manpower, sysadmins, hardware costs, electricity, heating, cooling, tech support, etc --AND-- given that they can buy bandwidth at $400 per megabit, --AND-- given the gamble that at any moment, only 10% of their DSL users will be maxing out their connection, the ISP stil has to charge at least $45 per month per 1.5 Mbit DSL user. More likely it will cost at least $90 before they even -begin- to bring in a profit.

Peer to Peer filesharing is a Big Thing and is growing at a huge rate, they're no denying that. With Napster alone many people are finding their unattended PC using almost half of their DSL thruput at any given moment.

There is no free lunch, your ISP, no matter how well connected, can't create bandwidth from nothing. Regardless if you get 5 Mbit access via your cable modem or can overclock your Pentium 4 to 1.8 GHz.

It has to happen

[Eoli]

Usage insensitive pricing of Internet access can support market development initiatives, particularly when relatively few players participate, each having made a significant commitment to lease or invest in transmission facilities. With the passage of time, more ISPs have entered the marketplace, often without the need for, or interest in making substantial investments in facilities. Later entrants may serve smaller geographical regions, and may have a deliberate strategy of "free riding" the facilities investment of other operators who still agree to accept traffic at quasi-public interconnection points. Likewise, because end user access to the Internet is typically priced on a low, flat-rated, "All You Can Eat" basis, no facility conservation incentive exists and therefore congestion can readily occur. As congestion threatens to impede quality of service, some ISPs have responded by prioritizing traffic streams, and by varying the price of network access on the basis of the transmission capacity and traffic volume of other ISPs seeking interconnection. This demand-based responsiveness soon might include reserved bandwidth that would provide higher service reliability and quality for a premium price. Resorting to traditional pricing mechanisms means parties causing congestion, or contributing comparatively less to congestion abatement, will incur higher costs of doing business. The responsible parties include smaller ISPs who lack the traffic, subscribership and transmission capacity needed to sustain highly reliable service in the face of increased demand and new Internet applications that require more bandwidth. Requiring payment for access to the facilities of other larger companies constitutes an efficient outcome, but one that likely will impose comparatively higher costs on smaller and rural ISPs and their subscribers.

This is good news.

[crucini]

The biggest barrier to consumer broadband has been the myth that you can get high-bandwidth, uncapped service for $40/month. Obviously, the people selling that myth were hoping that consumers would hardly use the bandwidth at all. When that turned out not to be true, profitability was threatened. Universities respond to this by trying to restrict the rights of their captive consumers. The crop of quotes in this article suggests that commercial ISP's, on the contrary, are seeing this as a legitimate usage that should be charged for. Ideally, I'd like to see them charge for transfer, rather than adopting a tiered scheme. But even the tiered scheme is a huge step in the right direction, away from the idea that "people who move lots of bits are troublemakers" towards the idea that "people who move lots of bits are our best customers".
This could have two wonderful effects. First, mega-corporations might find that it's more profitable to sell consumers transport, and remain content-agnostic than it is to build proprietary lock-in schemes and badly admin'ed caching proxies.
Second, of course, the US government tends to see only profit-making activities as legitimate. When the ISP's are profiting from p2p, they might serve as a counterbalance to the IP cartel that is currently 'educating' Congress on the evils of p2p.
Anyhow, charging per GB transferred is by far the healthiest business model because it gives ISPs incentives to upgrade their bandwidth, since pipes now show up as revenue producers to the bean counters. Anyone who sells 'bandwidth', on the other hand, is incentivized to minimize the customer's use of that bandwidth, whether by outages, restrictive AUP's, or other hassles.

The role will change

[autocracy]

ISPs will (hopefully) start moving away from the model where they try to provide everything they can on their homepage and will change focus to doing things like hosting e-mail and handing out IP addresses. Basically, the system will just be setup to do the basics and support (believe me, there will be incompetent Winblows users for YEARS to come that don't know how to do internet, and they WILL pay...)

ALL YOUR BASE ARE BELONG TO US!