Recipient-based Pay-By-Email
on
Replacing SMTP?
·
· Score: 2, Insightful
You're not correct. If the payment is charged by the email recipient, and the email recipient is willing to reject email that doesn't have attached payments (not universally true...), then ANYBODY can start pay-by-email system Right Now, and it'll protect them from unwanted email, just as anybody can implement "reply-to-my-turing-test-response" right now without anybody else being in control. Centralized systems are only needed if you want to force senders to pay non-economics-based amounts to somebody other than the recipient of the email.
Implementing it on a recipient-controlled basis is not only politically correct anarcho-capitalism (as opposed to central-planner-based and doomed to failure), it matches the main problem with spam, which is that it wastes the recipient's time, and only the recipient knows what that time is worth., and the recipient of the mail is the one who gets the money, not some middleman whose time isn't being wasted. If I dislike spam more than you do and am more willing to reject mail from people who might have interesting things to say, I can charge more money to read email than you do. Yes, email-provider ISPs also have higher workloads because ~50% of email is spam, but that's a much smaller amount of money per recipient than the wasted time, and they can charge the users accordingly.
Of course, if the recipient of the email gets paid for receiving the email, there's an obvious product for spammers to sell "Get paid receiving email";-)
Doing Hashcash wrong or correctly
on
Replacing SMTP?
·
· Score: 1
If you implement hashcash correctly, it forces spam senders to use a very large amount of CPU and therefore limits the rate at which they can send to (e.g.) thousands of messages per day instead of millions. If you implement it incorrectly, it forces spam senders to do the complex calculation once per version of their message, and lets them send it to millions of recipients. This does cut out some of the anklebiters, and makes them send "Dear Friend" messages instead of "Dear [recipient-name-here]", and helps Razor-like systems identify the spam, but it's easy for half-competent spammers and spamware-kit-sellers to automate around that.
The hashcash calculation has to be done per recipient to make the scaling work. This could be done in the message headers, or could be done in the envelope, but if you do it in the message headers, your mail system can't tell it's been done until it's accepted the message, while if you require it in the envelope or enforce it in the transfer protocol, it forces the sender to calculate it for every message. Adam mostly made the right choices.
However, if spammers can abuse open relays into doing their hashcash for them, then the scaling doesn't bother the spammers' machines, just the relay owners, and they're using hundreds of them in parallel so it's no problem. On the other hand, open relays aren't likely to implement new hashcash protocols, and mailers that do implement new hashcash protocols can be configured not to do relay by default.
Separate subdomains per user is fine for other reasons (if you've got your own subdomain with arbitrarily many email addresses, it's very easy to give every recipient a different one which lets you do lots of filtering tricks, such as having lots of honeypot addresses on your web site that catch spider-based and alphabet-search-based spammers.)
But it doesn't help the sending-from-lots-of-places problem much - you won't be able to send mail from a random cybercafe that way, so travelling users get screwed. (This happens increasingly often anyway - see some of John Gilmore's rants about having SMTP sent by his laptop get blocked by dialup-user blacklists.) On the other hand, you _could_ make things work more often by using tunnels to the ISP (ipsec, or SSL tunnels to the ISP's outbound mail relay) which would let your laptop use a consistent IP address from the ISP's space.
People may join the military or the Coast Guard with the intent of serving their countries, and the government abuses them by creating lots of publicity about how that's what their job is, and then uses them for entirely different jobs that advance the government's political agendas (such as stomping out Communism and supporting politically convenient dictators around the world) and in general endangers the American public by creating hostility against us (such as the recent Middle Eastern terrorism problems.)
Occasionally the Coast Guard is a military organization, keeping Nazi submarines from invading the US, and they have done a lot of really excellent work rescuing people from ship and boat accidents. Unfortunately, they've been largely turned into tax collectors and pirates, because they've been given the job of enforcing Prohibition (essentially all of the cargo container inspection is looking for drugs and import tax evaders, not to keep our bloody arses safe), and are authorized to attack ships that might be smuggling.
But unlike the Coast Guard, who have some good jobs as well as bad jobs, the people who "monitor land border crossings" are entirely bad - at best they're tax collectors (a few years ago the Customs department had the gall to be wearing uniform patches saying "US Customs Service - Defenders of Liberty") and at worst they're the thugs who want to keep immigrants with the wrong skin color from entering our country.
The airport security officials are primarily there to make the American public think that the Bush Administration is doing something to make us safer, and they've been interfering with civil liberties for years. Yes, they've probably kept a few wanabee hijackers from doing hijacking, but after the TWA Flight 800 explosion was shown to have been from mechanical causes, they didn't stop the unAmerican practice of demanding travel documents, and they've had a consistent policy of lying about policies, having secret policies, and encouraging the airlines to lie about policies.
Cost-effective ways to actually increase our security? Stop the US military and US foreign aid from supporting abusive regimes around the world - Saddam Hussein and Osama bin Laden aren't the first bad guys that we've provided military assistance to and then later gone to war against. Get rid of America's weapons of mass destruction. Do free trade with everybody.
That's "Put the cells in PARALLEL for increased lifetime." An external battery pack may be a more realistic option and I may do that.
It wasn't a successful purchase, but it wasn't a stupid one - the camera's about 1.5" square and very thin, the kind of pocket-sized device that's no trouble to carry around even if you don't expect to need it, was only $39, and my wife liked it so it was worth trying out.
The reason there's a huge market for $29 PC video cameras is because there's semi-adequate support for conferencing with them. MS Netmeeting is happier over broadband connections, and it's probably the most widely distributed H.323 application out there.
It's not the first or best low-end video conferencing software - CU-SeeMe from Cornell University is older, and does an adequate job even over 28.8kbps modem lines, though you get more frames per second at higher speeds. The early versions were free, and the later versions were commercialized by White Pine.
I'm assuming that by "PK" you mean "Public Key"? DES isn't public key, it's the standard symmetric secret-key algorithm, and crippled-DES is some variant on that.
Look, maybe those crazy Standards Committee Thugs like to use Coulombs, but here in the US, the battery packages list capacity in milliamp-hours, so that's the way to compare them.
Also here in the US, food energy is commonly measured in kilocalories, though all the dietary stuff tends to drop the "kilo" and call them calories. In some other places, at least including Australia, it's labeled in Joules, so I've seen "Low-Joule Cola":-)
I've seen these at 650-700mAh, but I'd really like something bigger. The AA sizes have gone from 1600 to 2200 in the last year or two, and it sure would be nice if the smaller size batteries could improve their lifetime. This is especially an issue for one of my digital cameras, which is a battery-burner that only takes AAA (and is too dumb to let you change batteries without losing all its pictures:-( Really cretinous design, which is especially annoying because it's a cute tiny camera. The thing takes 640x480 pictures and runs out of battery long before it runs out of memory.)
Cadmium is a mean nasty ugly toxic heavy metal, and everybody knows heavy metal went out with the 80s. However, some people have commented here that NiCd batteries are much better for high-current-drain applications, where you want to discharge the battery's power very quickly, e.g. toy airplanes and racing toy cars that want lots of power to go fast. You'll get less total milliamp-hours out of them, but you can get a lot more milliamps for a short period of time if you want.
Also, with laptop NiMH batteries, I've seen memory-effect-like problems. Not as bad as NiCd, but they do wear out. Using them on a twice-daily long train commute isn't the best thing for them, and I quickly learned that running the Great INternet Mersenne Prime Search in the background isn't something laptop batteries were designed for - the increased CPU load really makes a difference.
I have no idea if any of my chargers are -deltaV or trickle or what.
The RayOVac chargers can handle rechargeable alkaline, NiCd, or NiMH and cost about $10-15, and hold four AA or AAA batteries at once.
The GE/Sanyo charger cost $29 including two AAs, holds two AAA or four AA batteries, and was the only one in the store that said on the package that it could use 240 Volts (I needed one that would work in Europe with only a plug-adapter and no transformers.) Output 1.2V (so not useful for alkalines), 920mA/battery for two batteries or 460mA/battery for four batteries. Says it'll charge two NiCd AA 700mAh in ~1 hour, 2 NiMH AA 1700mAh in ~2 hours, twice as long for 4 batteries, a bit faster for AAA.
A 9-volt NiMH charger was about $10. I don't know if it's really 9V or 7.2V. I think the batteries were about $5 each.
Rechargeable alkaline batteries seem to cost about $1.30 for AA or AAA.
NiMH batteries cost about $1.80 for 10-packs of cheap AA, and about $2.20-2.50
NiCd batteries contain Cadmium, an evil dangerous toxic heavy metal, so I never buy them unless they're built in to products.
Book? I suppose there might have been one, but it was done as a movie by some ex-Disney animators. I didn't see that either, just saw the ads for it in the newspapers and probably a Siskel&Ebert review or something.
According to several other posters, it's running DES-crippled-to-40-bits as its encryption algorithm, and in ECB mode, not CBB, just to make it easier. Regular 56-bit DES took about a day for distributed.net to crack in ~1998 (though they got lucky - should have taken them ~2-3 days.) Since then, computers have gotten much faster, and this is 2**16 easier. (Technically that's only true if the crippled keyspace can be searched efficiently, like the full keyspace can, but that should be doable, and worst-case is no worse than single-DES.) True, the EFF machine in John Gilmore's basement hasn't gotten any faster, but it's been sitting there collecting dust for years, and somebody who wanted to spend another $250K to build a new one would get a much faster machine today - and if it's the RIAA, they could pay for it with the first couple of lawsuits against file-sharers.
Not only will it not keep government supercomputers out for weeks, it won't keep the RIAA out of your disk for weeks if they confiscate it. Besides, the RIAA can subpoena you to make you hand them the key dongle. Also, this is only useful against people who have physical possession of your disk when your machine isn't running - if your machine's running with the disk mounted, it's no different than a regular disk, so querying your Kazaa file-sharer will work just fine, or running a search program on your machine.
First of all, there's not a real shortage of IPv4 addresses - there are about 70 reserved Class A addresses that can handle another billion or so users if allocated carefully, plus whatever's left of Class B and Class C space and anything we invent for Class E.
But 6to4 tunnelling doesn't solve the IP address shortage. If your users can handle 6to4, they can probably handle NAT as well, and you can give them 10.x without bothering with IPv6 tunnels. If your users _can't_ handle NAT, because they need REAL IP ADDRESSES to give people in the real world, they probably need real IPv4 addresses, not IPv6 addresses.
IPv6 tunnels do help for customers of different ISPs who want to talk to each other without doing double NAT (10.x.x.x to Registered IPv4 to 10.y.y.y), so I suppose there's a bit of value there. On the other hand, *your* ISP doesn't need to offer it - if you can set up a tunnel to an ISP who's doing IPv6 tunnel brokering, that gives you real routes that other IPv6 users can use to reach you, even if your ISP is clueless. This means that for many users, IPv6 tunnel brokering is a service just like email is a service - it's convenient if your ISP offers it, but you can just as well get it from somebody else.
Y'all are confused about the real costs and prices of IP addresses. There's the cost of getting dedicated globally routable IP address space, and there's the cost of getting static IP addresses out of your ISP's space. Yes, many ISPs will charge you extra money for using a static IP address on your DSL, but that's mainly because it's more work for them to manage static than DHCP. (On the other hand, some ISPs force you to use PPPoE, because it's easier to shut off your account quickly using PPPoE, which would have also been easy if they used static...)
Globally routable blocks of IP addresses are the real constraint for the Internet, because every "backbone" router in the Internet needs to know the best routes to each block. There are currently about 100,000 of them, and calculating routing tables takes a lot of memory - it's only been recently that the biggest Cisco routers could calculate that many using BGP. Within each block, it's the block owner's problem to find the best route to the destination - the rest of the net doesn't need to care about that. The advantage of having your own routable address block is that you can take it with you when you change ISPs (so you don't need to renumber your network), and you can get service from multiple ISPs aat the same time to improve your reliability and shorten your routes to everywhere. There are a group of old grandfathered Class C/24 addresses that are globally routable, but otherwise you need about a/20 (for a long time it was a/19.)
Neither of these constraints on address space will be changed much by IPv6. There's some work being done on making it easier to summarize routes into larger blocks while preserving routing efficiency, but I'm not particularly convinced it's getting anywhere (though that was certainly a goal - for now we're solving the problem by watching Moore's Law make memory much cheaper). The amount of work required to manage a customer doesn't change any. Address space is almost infinitely large - instead of giving you one static IP address, your ISP should easily be able to give you a/64 which has 2**64 addresses in it (roughly 2**16 subnets of 48 bits, so you can use your Ethernet MAC address to autoaddress IP instead of really needing DHCP, kind of the way Netware worked). It'll be a bit easier for people to manage their own subnet space because a/48 gives you 2**16 subnets of/64, so nobody in the world should be short on the stuff, but it's not a big help compared to the routable address issue.
There are a few exceptions, like MIT, but many of the universities like Stanford that used to have Class A addresses have given them back. (There are also a few big ISPs that are dead now and could give them back.) If you go check out the registries, you'll find that there really are a lot of Class A networks that have been given back or were never given out or are otherwise "reserved" for future use - there are about 70 free. Class B and Class C space are tighter.
If the IP-enabled cell phone business takes off to the point that they need 10-20 Class A addresses to handle the 160-320 million new customers, all of us in the phone industry will be deliriously happy and we'll have the money to spend converting things to IPv6. (For a while, it's more likely the phones will run IPv6 with IPv4 NAT or web proxies to handle the portions of the network that don't need full-time IPv4 connectivity.)
The other big potential use is if a Billion extra people get full-time internet connections. The main reason this is likely to happen eventually is if third-world telephony deployment is done with IP telephony instead of traditional circuit phone switches, which the wired parts of the network may very well be. But that's going to take a while - and it'll take getting rid of telephone monopolies for most of those billion people, which may take some time.
No, most ISPs don't have AUPs banning servers. Unfortunately, most cable modem companies do, and some big DSL providers also do, because they're suicidally clueless about where the next killer application that gets people to buy broadband will emerge from - but even then, they're pretty fuzzy about what a "server" is - an IM chat client is also a server, and a game that has user-to-user chat or other communications is also a server, but they know their users want those things so they allow them.
Most dial ISPs don't ban servers, though most of them don't give you 7x24 use of a static IP address, which makes running a server somewhat difficult, and most DSL services also tend to do dynamic addresses. But you can work around that.
It's because 127.*.*.* is logically a Class A address, under the old classful addressing scheme, so it gets a whole/8 whether it needs it or not. There's no good reason for it to be a/24 - a/32 would make some sense, but anything between that and/8 is artificially silly. On the other hand, if you've got any application that needs more than two loopback addresses, there's no reason to expect it not to use MAX_SOCKETS_PER_PROCESS or MAX_PROCESSES_PER_MACHINE number of loopbacks if it's in the mood.
The main real-world applications I've seen for 127.*.*.* values with host-parts greater than 1 has been spam-blocker DNS responses that have to return something that's syntactically an IP address but is easily distinguished from a valid IP address - many of them return values from 127.0.0.2 to 127.0.0.6 depending on what reason you might have for not wanting to talk to some-machine.example.net.kr. I've personally found them to be useful mnemonics for host-file entries for machines I don't want to talk to ("127.0.0.2 doubleclick.net" makes their cookie requests not go through...).
Surprisingly, on my Win2K machine, pinging 127.255.255.255 didn't respond, though 127.2.3.4 did....
At least here in Silicon Valley, there's an important etiquette details about product-related T-shirts, which is that if it's an interesting shirt, people will ask you about it so you should have an answer ("Oh, did you work on the FooBar 6? Those were cool!" "No, but we were a beta customer for it.")
Levis convinced the business world that Dockers should be the Official Business Casual dress code, replacing the wool suit, which gradually meant that black jeans are as formal as I ever need to get for customer visits. (Yay! Cotton is much more comfortable.) I normally wear long-sleeved shirts anyway, so that wasn't an issue, and a good white cotton shirt handles the formality bit when that's appropriate.
San Francisco is a bit more formal than Silicon Valley, not so much because they have more bankers than programmers but because the weather forces you to wear a jacket half the year anyway, but most of the bank employees are IT people so you don't need the suit&tie anyway.
Yes, there are some places that expect you to wear specific costumes, but normally if you dress neatly the ponytail isn't a problem, especially in a technical job. I didn't grow my beard until after I'd already done the job interview, but that was just a timing thing, not a strategy, and business environments were more formal 25 years ago. It was never a problem, though I did trim it shorter when I did a job interview 10 years ago.
As one customer of mine said "Extra points because it's a Jerry Garcia tie, but we don't wear ties here so you shouldn't either..."
OK, so don't wear Tevas, wear the formal black-colored Birkenstocks and socks, not the informal natural-colored ones. You'd be surprised what's acceptable in most environments. I keep a pair of wingtip salesman shoes in my car for visiting more formal customers, but wear the birks when I'm talking to engineers.
Obviously this doesn't apply if you're working somewhere that you need closed-toed shoes for protection from dangerous objects...
I liked MS Barney. (OK, I didn't buy one, but...) It was so blatantly over-the-top evil that you simply had to admire them for coming out with the product.
Survival Research Labs does shows of large robots bashing each other up, with fire, much crashing and bending of metal, etc. They did a show near my office back in the boom (rented a parking lot under the freeway). One of their destructive robots had a MS Barney on top. I don't know if he was animated or just duct-taped on, but he was both a Hood Ornament of Evil and a good target for the other robots.
I'm not that bothered by MS's questionable try-to-leverage-your-dominant position tactics. That's business, and if you didn't like it at any point since 1984, you could have bought a Macintosh. I don't care that they bully PC makers into bundling their products - that's a business choice, and you can buy from Mom&Pop shops instead. I don't care that they might have been violating US or EU anti-trust rules - anti-trust rules are bogus, and the US and EU are also big evil bloated bureaucracies; I'd rather deal with MS and IBM.
I dislike MS because so many of their products have been so *shoddy*, unreliable, crashing, poorly documented, anti-standardization, failing to take advantage of technologies that have been well known to the public from years before, bloated and slow. The one questionable business/marketing tactic of theirs that does annoy me is that each version of their popular products is deliberately incompatible with previous versions in ways that force you to upgrade if you're going to keep communicating with other users. Some examples are naming the "Program Files" directory in a way that broke all previous well-behaved software installation programs, forcing software vendors to either dual-version or only support the newer OS, and making each version of Word have something that makes new-version documents usually fail on old Word versions and old documents get weird things done to them on new versions.
On the other hand, there were the MS Barney Dolls, which were so over-the-top evil that you just had to admire them....
Hey, even though they've got a few orders of magnitude variation in the costs, at least they're talking about the costs of wasted employee time, which is the real cost, and not whining about bandwidth usage. Yes, when I'm on dialup, downloading spam takes some time, but on a work LAN or on DSL at home it's very little download time, and the bits it consumes are usually a lot less than reading Slashdot.
That's also true for ISPs - web traffic carries a lot more bits than spam, and while spam email probably outnumbers real email by byte-count as well as message count, it's not really a big deal for connectivity-provider ISPs. (Email-specialist ISPs are obviously another case entirely.) On the other hand, the worker-time cost of handling spam complaints and trying to keep filters up to date is more important than the cost of the bits.
Almost all email programs let you display the sender and subject without opening the message. 90% of the time it's pretty easy to tell just from reading those whether to delete it without opening - filters can often trash many of those messages automatically, but they can also speed up the decision time by marking suspicious messages.
Slashdot Mirror
Implementing it on a recipient-controlled basis is not only politically correct anarcho-capitalism (as opposed to central-planner-based and doomed to failure), it matches the main problem with spam, which is that it wastes the recipient's time, and only the recipient knows what that time is worth., and the recipient of the mail is the one who gets the money, not some middleman whose time isn't being wasted. If I dislike spam more than you do and am more willing to reject mail from people who might have interesting things to say, I can charge more money to read email than you do. Yes, email-provider ISPs also have higher workloads because ~50% of email is spam, but that's a much smaller amount of money per recipient than the wasted time, and they can charge the users accordingly.
Of course, if the recipient of the email gets paid for receiving the email, there's an obvious product for spammers to sell "Get paid receiving email" ;-)
The hashcash calculation has to be done per recipient to make the scaling work. This could be done in the message headers, or could be done in the envelope, but if you do it in the message headers, your mail system can't tell it's been done until it's accepted the message, while if you require it in the envelope or enforce it in the transfer protocol, it forces the sender to calculate it for every message. Adam mostly made the right choices.
However, if spammers can abuse open relays into doing their hashcash for them, then the scaling doesn't bother the spammers' machines, just the relay owners, and they're using hundreds of them in parallel so it's no problem. On the other hand, open relays aren't likely to implement new hashcash protocols, and mailers that do implement new hashcash protocols can be configured not to do relay by default.
But it doesn't help the sending-from-lots-of-places problem much - you won't be able to send mail from a random cybercafe that way, so travelling users get screwed. (This happens increasingly often anyway - see some of John Gilmore's rants about having SMTP sent by his laptop get blocked by dialup-user blacklists.) On the other hand, you _could_ make things work more often by using tunnels to the ISP (ipsec, or SSL tunnels to the ISP's outbound mail relay) which would let your laptop use a consistent IP address from the ISP's space.
Occasionally the Coast Guard is a military organization, keeping Nazi submarines from invading the US, and they have done a lot of really excellent work rescuing people from ship and boat accidents. Unfortunately, they've been largely turned into tax collectors and pirates, because they've been given the job of enforcing Prohibition (essentially all of the cargo container inspection is looking for drugs and import tax evaders, not to keep our bloody arses safe), and are authorized to attack ships that might be smuggling.
But unlike the Coast Guard, who have some good jobs as well as bad jobs, the people who "monitor land border crossings" are entirely bad - at best they're tax collectors (a few years ago the Customs department had the gall to be wearing uniform patches saying "US Customs Service - Defenders of Liberty") and at worst they're the thugs who want to keep immigrants with the wrong skin color from entering our country.
The airport security officials are primarily there to make the American public think that the Bush Administration is doing something to make us safer, and they've been interfering with civil liberties for years. Yes, they've probably kept a few wanabee hijackers from doing hijacking, but after the TWA Flight 800 explosion was shown to have been from mechanical causes, they didn't stop the unAmerican practice of demanding travel documents, and they've had a consistent policy of lying about policies, having secret policies, and encouraging the airlines to lie about policies.
Cost-effective ways to actually increase our security? Stop the US military and US foreign aid from supporting abusive regimes around the world - Saddam Hussein and Osama bin Laden aren't the first bad guys that we've provided military assistance to and then later gone to war against. Get rid of America's weapons of mass destruction. Do free trade with everybody.
It wasn't a successful purchase, but it wasn't a stupid one - the camera's about 1.5" square and very thin, the kind of pocket-sized device that's no trouble to carry around even if you don't expect to need it, was only $39, and my wife liked it so it was worth trying out.
It's not the first or best low-end video conferencing software - CU-SeeMe from Cornell University is older, and does an adequate job even over 28.8kbps modem lines, though you get more frames per second at higher speeds. The early versions were free, and the later versions were commercialized by White Pine.
I'm assuming that by "PK" you mean "Public Key"? DES isn't public key, it's the standard symmetric secret-key algorithm, and crippled-DES is some variant on that.
Also here in the US, food energy is commonly measured in kilocalories, though all the dietary stuff tends to drop the "kilo" and call them calories. In some other places, at least including Australia, it's labeled in Joules, so I've seen "Low-Joule Cola"
I've had no problem finding them in Fry's, but maybe you don't live near one. Check out Lenmar.com - it's their "Charge*ables" brand.
I've seen these at 650-700mAh, but I'd really like something bigger. The AA sizes have gone from 1600 to 2200 in the last year or two, and it sure would be nice if the smaller size batteries could improve their lifetime. This is especially an issue for one of my digital cameras, which is a battery-burner that only takes AAA (and is too dumb to let you change batteries without losing all its pictures :-( Really cretinous design, which is especially annoying because it's a cute tiny camera. The thing takes 640x480 pictures and runs out of battery long before it runs out of memory.)
Also, with laptop NiMH batteries, I've seen memory-effect-like problems. Not as bad as NiCd, but they do wear out. Using them on a twice-daily long train commute isn't the best thing for them, and I quickly learned that running the Great INternet Mersenne Prime Search in the background isn't something laptop batteries were designed for - the increased CPU load really makes a difference.
Book? I suppose there might have been one, but it was done as a movie by some ex-Disney animators. I didn't see that either, just saw the ads for it in the newspapers and probably a Siskel&Ebert review or something.
Not only will it not keep government supercomputers out for weeks, it won't keep the RIAA out of your disk for weeks if they confiscate it. Besides, the RIAA can subpoena you to make you hand them the key dongle. Also, this is only useful against people who have physical possession of your disk when your machine isn't running - if your machine's running with the disk mounted, it's no different than a regular disk, so querying your Kazaa file-sharer will work just fine, or running a search program on your machine.
But 6to4 tunnelling doesn't solve the IP address shortage. If your users can handle 6to4, they can probably handle NAT as well, and you can give them 10.x without bothering with IPv6 tunnels. If your users _can't_ handle NAT, because they need REAL IP ADDRESSES to give people in the real world, they probably need real IPv4 addresses, not IPv6 addresses.
IPv6 tunnels do help for customers of different ISPs who want to talk to each other without doing double NAT (10.x.x.x to Registered IPv4 to 10.y.y.y), so I suppose there's a bit of value there. On the other hand, *your* ISP doesn't need to offer it - if you can set up a tunnel to an ISP who's doing IPv6 tunnel brokering, that gives you real routes that other IPv6 users can use to reach you, even if your ISP is clueless. This means that for many users, IPv6 tunnel brokering is a service just like email is a service - it's convenient if your ISP offers it, but you can just as well get it from somebody else.
Globally routable blocks of IP addresses are the real constraint for the Internet, because every "backbone" router in the Internet needs to know the best routes to each block. There are currently about 100,000 of them, and calculating routing tables takes a lot of memory - it's only been recently that the biggest Cisco routers could calculate that many using BGP. Within each block, it's the block owner's problem to find the best route to the destination - the rest of the net doesn't need to care about that. The advantage of having your own routable address block is that you can take it with you when you change ISPs (so you don't need to renumber your network), and you can get service from multiple ISPs aat the same time to improve your reliability and shorten your routes to everywhere. There are a group of old grandfathered Class C /24 addresses that are globally routable, but otherwise you need about a /20 (for a long time it was a /19.)
Neither of these constraints on address space will be changed much by IPv6. There's some work being done on making it easier to summarize routes into larger blocks while preserving routing efficiency, but I'm not particularly convinced it's getting anywhere (though that was certainly a goal - for now we're solving the problem by watching Moore's Law make memory much cheaper). The amount of work required to manage a customer doesn't change any. Address space is almost infinitely large - instead of giving you one static IP address, your ISP should easily be able to give you a /64 which has 2**64 addresses in it (roughly 2**16 subnets of 48 bits, so you can use your Ethernet MAC address to autoaddress IP instead of really needing DHCP, kind of the way Netware worked). It'll be a bit easier for people to manage their own subnet space because a /48 gives you 2**16 subnets of /64, so nobody in the world should be short on the stuff, but it's not a big help compared to the routable address issue.
If the IP-enabled cell phone business takes off to the point that they need 10-20 Class A addresses to handle the 160-320 million new customers, all of us in the phone industry will be deliriously happy and we'll have the money to spend converting things to IPv6. (For a while, it's more likely the phones will run IPv6 with IPv4 NAT or web proxies to handle the portions of the network that don't need full-time IPv4 connectivity.)
The other big potential use is if a Billion extra people get full-time internet connections. The main reason this is likely to happen eventually is if third-world telephony deployment is done with IP telephony instead of traditional circuit phone switches, which the wired parts of the network may very well be. But that's going to take a while - and it'll take getting rid of telephone monopolies for most of those billion people, which may take some time.
Most dial ISPs don't ban servers, though most of them don't give you 7x24 use of a static IP address, which makes running a server somewhat difficult, and most DSL services also tend to do dynamic addresses. But you can work around that.
The main real-world applications I've seen for 127.*.*.* values with host-parts greater than 1 has been spam-blocker DNS responses that have to return something that's syntactically an IP address but is easily distinguished from a valid IP address - many of them return values from 127.0.0.2 to 127.0.0.6 depending on what reason you might have for not wanting to talk to some-machine.example.net.kr. I've personally found them to be useful mnemonics for host-file entries for machines I don't want to talk to ("127.0.0.2 doubleclick.net" makes their cookie requests not go through...).
Surprisingly, on my Win2K machine, pinging 127.255.255.255 didn't respond, though 127.2.3.4 did....
Levis convinced the business world that Dockers should be the Official Business Casual dress code, replacing the wool suit, which gradually meant that black jeans are as formal as I ever need to get for customer visits. (Yay! Cotton is much more comfortable.) I normally wear long-sleeved shirts anyway, so that wasn't an issue, and a good white cotton shirt handles the formality bit when that's appropriate.
San Francisco is a bit more formal than Silicon Valley, not so much because they have more bankers than programmers but because the weather forces you to wear a jacket half the year anyway, but most of the bank employees are IT people so you don't need the suit&tie anyway.
As one customer of mine said "Extra points because it's a Jerry Garcia tie, but we don't wear ties here so you shouldn't either..."
Obviously this doesn't apply if you're working somewhere that you need closed-toed shoes for protection from dangerous objects...
Survival Research Labs does shows of large robots bashing each other up, with fire, much crashing and bending of metal, etc. They did a show near my office back in the boom (rented a parking lot under the freeway). One of their destructive robots had a MS Barney on top. I don't know if he was animated or just duct-taped on, but he was both a Hood Ornament of Evil and a good target for the other robots.
I dislike MS because so many of their products have been so *shoddy*, unreliable, crashing, poorly documented, anti-standardization, failing to take advantage of technologies that have been well known to the public from years before, bloated and slow. The one questionable business/marketing tactic of theirs that does annoy me is that each version of their popular products is deliberately incompatible with previous versions in ways that force you to upgrade if you're going to keep communicating with other users. Some examples are naming the "Program Files" directory in a way that broke all previous well-behaved software installation programs, forcing software vendors to either dual-version or only support the newer OS, and making each version of Word have something that makes new-version documents usually fail on old Word versions and old documents get weird things done to them on new versions.
On the other hand, there were the MS Barney Dolls, which were so over-the-top evil that you just had to admire them....
That's also true for ISPs - web traffic carries a lot more bits than spam, and while spam email probably outnumbers real email by byte-count as well as message count, it's not really a big deal for connectivity-provider ISPs. (Email-specialist ISPs are obviously another case entirely.) On the other hand, the worker-time cost of handling spam complaints and trying to keep filters up to date is more important than the cost of the bits.
Almost all email programs let you display the sender and subject without opening the message. 90% of the time it's pretty easy to tell just from reading those whether to delete it without opening - filters can often trash many of those messages automatically, but they can also speed up the decision time by marking suspicious messages.