I forget how long I've been in this game...

My time frame from memory is about 1994. I came to Minneapolis in 93, and spent roughly a year looking for an ISP. MRNet was the only real
provider, and as a consortium, they priced service assuming you were a
reseller. A friend of mine at 3M actually did use them, and he was
paying $185/mo for dialup, well beyond what I could afford.

Not to surprisingly, this was a reasonable rate from a reseller
perspective. The going rate for a T1 of internet was in the $5K range.

Roughly a year later, some internet enthusiasts put together the
wonderfully named "winternet" and offered access for $25/mo. I was
absolutely thrilled!

They were, of course, selling at a rate that reflected the natural
over subscription of casual internet use, not the potential cost of full
usage. They saw what ISPs were to be before others did.

Last I checked, wholesale reseller bandwidth was around $100/Mb. That's the rate they charge each other. Here in Minneapolis, the ILEC will sell you a T1 of internet for around $300-$500. If you need guaranteed bandwidth, that's a pretty reasonable markup considering it includes transport cost in addition to internet.

If you are paying $30/mo for "7Mb" (the most aggressive current local consumer pricing) you are pretty much in the position of buying dollar
bills for nickels, nice if you can get it, but clearly there's a "catch".

I remember hunting for an ISP back in the mid 90's. All ISPs priced there service as if bandwidth was going to be 100% utilized. A cheap rate was roughly $200/month...

Overselling bandwidth is a good deal for both the provider and the consumer. Without it the net as we know it would have been stillborn. Yes there are abuses, but the alternative is far worse.

In some more perfect world, an ISP could be counted on to clearly explain all the tradeoffs, but in the world I live in, marketeers speak to rubes, and ISPs differentiate themselves via specious and irrelevant shiny talk of "7MBS bandwidth"

The "harm" you experience when the ISP can not fully deliver in return for the artificially low amount you are spending doesn't really hold much weight. If you need the bandwidth, there are many who will honestly sell it to you. It's just that the real premium for that is 2X to 10X the shared rate.

2 examples....

1) I went to visit my grad student mother 5 years ago in New Hampshire. On a whim, I went to the U library and looked at the computers. I looked up a relatively obscure 18th century figure I'm interested on the library catalog. There, on the catalog system were digitized copies of small run monographs that were only really available in a few british libraries. At the time I was floored.

2) In the context of something totally different, I became fascinated by the role of Mark Twain as blurring what "honesty" means. About 15 minutes of googling came up with a Google books reference i would likely have never found otherwise that spoke in a rich and direct way to my thoughts. I ended up buying a used copy of the book. It was an academic book, and I would not have been a likely candidate for purchasing it (I couldn't justify the $60 it would have cost to buy new).

I love books. I grew up with books. Kindle may be getting there, but books are a great form factor. That said, books are still just a medium. The message is what really counts. Putting culture online has many wonderful and far-reaching effects. It also is, and will continue to create a sea-change which will undoubtedly hurt people.

Can you provide any actual cases even remotely resembling this? I do not believe you can, and thus i believe your argument carries almost no weight.

The settlement specifically applies to works where the intent of the copyright owner is not discoverable. Your example seems to have no application here.

Authors wishes raise some interesting questions. Some that seem worth mentioning.

1) Kafka explicitly did requested his work not be published. Should we honor his wishes? This is not an uncommon situation.

2) At what point do the authors wishes expire? One of the central goals of copyright is to expire that right. Given that the works in question are all quite old, and that the probability that the author has expired, what credence should we give that authors wishes.

3) Since the original book can be resold, and viewed by non-owners (i.e. library patrons) and the right to control that is explicitly denied the author, what distinguishes the Googles attempts?

It's easy to make too much of cleartext SMTP. Though I can think of attack vectors that would enable wire sniffing attacks, most require significant knowledge of some middleman network architecture. Pretty much all SMTP traffic is traveling over switches, and non-trivial to tap.

Compromising the mail server is far more fruitful. At a minimum the cleartext SMTP traffic become much more accessible. Wire sniffing is passive, and therefore a little scarier, but server compromise is likely much more common, and more dangerous.

Actually, Google's main source of revenue is AdSense.

True, but that does almost nothing to weaken the GP's point. Without the search engine spidering and the enormous amount of search data to mine, adwords/adsense has no real value. It's effectiveness is almost completely dependent to the visible, and very popular search engine.

I worked for an ISP that provided service to hotels. VPN configs were the major source of problems. We implemented a captive portal to try to smooth over issues like

SMTP rejection (SMTP-AUTH was not common, the portal provided silent redirect to local mail server)

Accountability/Abuse -- The rooms were hard-wired, and captive portal gave us some retroactive sense of what room was generating abusive traffic.

Splash-screen/terms-of-service

DNS redirection is one of the core techniques for establishing captive portals. I rather doubt that many smaller ISPs are doing the "sponsored link" DNS redirect. Maybe things have changed since I left, but I suspect there is no significant benefit and some real cost involved for sponsored redirects for all but the largest ISPs.

Most of the support calls were over VPN software. Since all traffic was redirected until the splash screen was agreed to, a small but significant segment of VPN client configs broke. I very much suspect that is the real source of the initial posters issues.

Hash functions are a general comp-sci idea, of which crypto hashes represent a very small subset of applications

There is no requirement that a general hash function is one way.

It's a variation on an identity hash, where the hash space and the key space are the same size (one to one mapping).

It might not be a particularly useful hashing function....

A little quick googling confirmed my suspicions. Mr Clarke donated books that were sent without return postage

Now why is it part of the authors duty to spring for time and postage to mail a book signing request back to the requestor?

I'd say the sender is thoughtless and selfish, regardless of what you might think of Mr Clarke's behavior.

"Joe" seems guilty of the same disease many programmers are, premature optimization. This is a perennial issue with coders, not something confined to greybeards.

I've seen ASP developers worry about the cost of function calls, and use that as a rationale for write only, unreusable and barely maintainable code. I've seen brand new VB coders worry about "large" arrays. I've seen perl coders tell me not to make function calls as it "slows down the code". These are all personal examples, fill in your own as appropriate.

"We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil."

Donald Knuth

It's as applicable today as it ever was.

I essentially agree with you about the main post. If there shipping cable TV, much of the cost is already sunk.

I was responding to your pricing quotes, for two reasons:

1. The quote is a lowball quote. The cable TV operation might well be able to get that pricing, but it's unlikely. Why not offer a more realistic quote. I'm pretty sure the numbers work out similarly (i.e. the cable folks should spring for more bandwidth).
2. The lowball quote feeds the "I'm paying for all that bandwidth all the time for my $50" crowd. It creates the impression that consumer bandwidth should be backbone quality at $50 a month.

I probably should have picked the post I responded to a with a little more care, as the "ISPs are evil" post was not the direct parent to yours. I saw that post, and saw yours shortly after with the lowball quote.

Before you take offense, I should say that $50/Meg does fit with my impressions of wholesale price for internet, so the quote is not entirely unreasonable, but it is certainly a come-on pricing that, in standard Salesmen-pitch style creates the impression that the true cost is much cheaper then it truly is.

I agree that 70x over-subscription is likely too high, and the request
makes it clear that it is to high, but your pricing
estimates are too low to guide any discussion.

A quick Google search for "DS-3 pricing" shows a full DS-3 for $2200. That means 45 Mbps, so overselling that based on 5-year-old usage data means an ISP could sell 1080 Mbps. Across 400 customers, that's 2.7 Mbps each for $2,200 a month.

That's internet bandwidth pricing, which is not the most expensive
piece, oddly enough. Nor, for a small operation, is it the main
component in pricing.

What does it take to actually implement this?

Well

• A port charge. You are usually charged rent for using up the
  physical port. With the lowball quote you gave, the provider is almost
  certainly going to hide some charges here
• A datacenter. Depending on how well done this is, the costs can vary
  widely. It's clear the original querier's operation is lowball, but even
  a mediocre data center has significant cost associated with it. Think
  power, UPS, Generator, Air-conditioning, rent... Doing this right will
  be the major cost, even ignoring buildout.
• Loop charge. This is likely to be significantly more expensive then
  the bandwidth. Unless you are right next to the provider, the charges
  are high. In no way will it be less then $2200, and could easily top
  $10,000
• buildout. Typical DS3 trenching charges will run you $100,000 -
  $300,000, payable up front. DS3 capable router will run you at list
  $5K, The telco will usually lease you termination equipment for a few K,
  or you could purchase, adding another 100K. Most of this is one-time,
  but that still needs to be factored into the total cost, and it can be
  very substantial
• support staff. Even a minimal staff will cost a whole lot more then
  the DS3

Running an ISP is not a way to get rich.

It bothers me when I read the grandparent post implying that ISPs rip
people off by not selling them dedicated bandwidth. Dedicated is an
order of magnitude more expensive. I remember trying to get access under
models that harkened closer to dedicated access, and it was bloody
expensive, I remember a $300/mo pricing for dialup.... Oversubscription
is good for all of us. The provider in question is not doing the right
thing (ordering more bandwidth) but it bothers me to see un-hinged rants
(grandparent post) suggesting that dedicated bandwidth is a reasonable
expectation.

WSH/vbscript is a fine thing. It let me feel somewhat at home in the last job I had with strong windows server work.

That said, it has an extremely important weakness, no way to implement recursive libraries. For one offs, it's just fine, but for anything more complex, the P's are a much better choice.

I did come across a complete hackish way to implement this, read the library in and interpret it. That's a pretty dangerous substitute for having real library support.

For what it's worth, the IIS vbscript does not have that weakness. It does allow libraries to call libraries.

It seems a real stretch to suggest that the numbers of people installing a hacked OSX would be more then a small fraction of either:

• those who leave the default install alone
• Those who install a pirated XP

I'd venture that most of the slashdot crowd would install there favorite OS. I have no real feel for what the average Dell customer would do though, and I rather suspect that few of us here do.

It is credible to me that a significant portion of the 32% is XP installs, but it also seems likely to me that over half of that 32% remain a linux install. As a prior post indicated, the price point is not huge, which limits the the re-installs to the fully intentional pirates (i.e. you are fully intending to pirate when you purchase). I rather doubt the few bucks saved means that much to most Dell customers, and that most of that 1/3 at least intend to use Ubuntu when they purchase.