I'll take TFA's word for the only-one-in-200-gets-kept figure (99.5%). We don't really know what fraction of those names are making _some_ money, because the 199 duds include some names that only get tasted once and some names that get repeatedly kited every time their 5-day period runs out, and the article doesn't really differentiate.
The cost of those 199 dud attempts is currently near-zero (so it's worthwhile kiting names even if you don't put much effort into predicting their quality very well, because you'll find out soon enough which ones work.) (Techically the price is slightly above zero, because of the cost of money on $6.42 for five days, so it's around $0.01 per name.) So the name-kiters can keep rolling over names today, either buying any name that makes over $6 or continuing to kite names for free while hoping that other DNS-abusers don't steal them at the end of the grace period (which they'd do for names that make The proposed ICANN fee raises that to $0.20 each, which is where the $46.22 figure comes in, basically $6 for the name you keep and $40 for the duds. That one seems reasonable - you'd still keep any name that makes more than $6.42/year in revenue, but you need to *average* over $46.22 per good name if you're going to stay in business because the duds cost you money. Also, the 20-cent fee for names turns over >73 times/year, which is ~$14.60, so you can't do the continuous-kiting game any more - if a name's not making $6/year, you drop it, and that'll probably cut down a lot on the transaction rates. (Again, that's not really precise, because even some of the duds will make non-zero money - $6/year is about 9 cents per 5-day-grace-period, but you might get some names making 5 cents during the period even if they're not making enough to actually buy.)
But the $1284/year = 200 * $6.42 is different - that's the cost of actually buying all 200 names, as opposed to buying the one good one and only paying the ICANN tasting fee on the duds. It's an even fuzzier figure - it really means that domain-tasting gamblers need to average $6.42/name to stay in the tasting business (which we're guessing won't happen), but again, some of those names might make $3 even if they're not making $6.42, so the best name doesn't need to make the whole $1284.
I would have used a longer term than "Bad", such as "Egregious" or "OverReaching" or "Abusive", but/. thought the title was too long....:-)
Maybe you haven't been watching the domain name policies evolve, and assume that they way they are today is both good and necessary. I've been paying attention to them since the mid-80s, and back then it wasn't clear that the market would even accept a system that had a single Root in charge of who could use what names - after all, UUCP worked quite well without it, and the Plan 9 folks had a bunch of really strong arguments for why local-centric naming was better. And I watched the IETF Ad-Hoc Committee work on policies for how to expand the TLD space while the Trademark Gods tried to prevent that from happening (and the bad guys won...) Those "obligations" you're referring to are rules ICANN made, and they'd have been controversial at the time except that ICANN had a policy of ignoring anybody who disagreed with them so controversy wasn't permitted.
The purpose of whois contact information was originally so that if something wasn't working, you could contact the owners of a name to fix it, and once they started charging money for domain names, it was also so that the registrar could reach a name registrant to tell them to pay their bill. That's all - those functions don't require that you provide your True Name, ICBM address, retina print from remaining eyeball, mother's maiden name, first born child's DNA, etc., and they can take care of the billing issues by accepting payment in advance or billing you by email.
From the standpoint of the Trademark Gods, which is what ICANN requires all their registrars to collaborate with, a DNS registration needs to have enough information to get you a subpoena with your name on it in case of a trademark lawsuit (which is where most of the requirement for True Name, Jurisdiction, and Postal Address came from). Various flavors of cops, spammers, and other anti-privacy types think this is just a fine idea, and have tried to extend it. On the other hand, some of the registrars have been providing some private registration services, in some case with some jurisdictional diversification. If you want to spend ~$100/year, you can also get your privacy the old-fashioned way - I've tracked at least one spammer's whois registration to the street address of "The Company Corporation", who've been the canonical business setting up cheap Delaware corporations for over a century, so even if I had successfully sued the spammers for big bucks, their only real assets were a file-folder in a drawer at The Company Corporation, and they could get back in business by spending another $100 for another corporate shell.
Obviously you can't do the same level of dispute resolution or problem-solving if you don't have somewhat usable contact information. It's really helpful to have the technical contact information for a domain be something other than an email address at that domain, because if the DNS is broken, you can't send them the email, but example-dot-com-technical-contact@yahoo.com is perfectly reasonable, and I'd rather not put my home or cellular phone number on a whois page that somebody could call at any time day or night (as opposed to an office phone, which would be perfectly reasonable for a domain owned by a business.) But if there's a legitimate domain name ownership dispute, and the registrar can't reach you by any of the contact mechanisms you choose to provide to them, it's reasonable for the registrar to resolve the dispute by saying you lose, and making it your problem to get the domain name back. Even that's got problems, especially for a personal domain - if you're on vacation for a month visiting your family in India, you might have trouble getting your email, but you also aren't going to see that paper subpoena that the alleged trademark owner put in your home mailbox.
And of *course* you were asking for ICANN to police people
As much as some people in the US government, and a few people like you in the Enemies-of-Evil business, would like it, ICANN shouldn't be in the censorship business, and for the most part they aren't. They're in the business of administering name space, and they're trying to make sure they don't sell names that violate existing trademarks - and even that's controversial, because there are businesses in different geographical areas with similar names (Joe's Pizza in different cities), and businesses in different types of business with similar names ("Coke" is that stuff you make by heating coal, and there's no trademark conflict with the brown liquid or the white powder), and the extent to which ICANN insists on violating the privacy of DNS name holders to make trademark lawsuits easy is pretty significant. For the most part, ICANN's dispute resolution process is designed to keep the registrars from having to go to court to deal with Dispute Resolution, because selling cheap domain names means they can't afford to do that, and ICANN doesn't want to pay for it either.
If you want to call your domain I-Got-This-Name-First.com, and somebody else already has that name trademarked, ICANN's policies should prevent you from getting it second, though arguably they could leave room in the DNS structure for disambiguating legitimate multiple uses of names. But if you want to call your website Pirated-Microsoft-Software.com, that's an issue between you and Microsoft to resolve the trademark conflict, and it's certainly not ICANN's business to decide if you really _do_ sell pirated Microsoft software.
Remember that judge who let those Cayman bankers get a temporary injunction against Wikileaks.org's domain name? Even he realized that it was a mistake, in spite of the number of people who don't like Wikileaks posting things that they don't like.
Currently, ICANN takes a 20-cent cut of the registration fee, the registry and registrars get to split $6.22 somehow, and the registrars get to charge whatever they like above that for convenient friendly service etc.
Back when somebody at ICANN invented the Annoying Grace Period that facilitates Domain Tasting, they mandated that the registry and registrars give the money back if somebody returns a domain name within 5 days. I guess it "seemed like a good idea at the time", but of course it's become obvious since then that that was a bad idea.
Think of the 20-cent fee as the beginning of ICANN saying "Yes, we made a mistake" - there's too much ego involved, and too many of the current crop of registrars making a profit on this scammery that they don't want to let go of, for ICANN to simply fix their mistake in one step, because now they've got to compromise with all that political pressure. So they're imposing the policy with their part of the fee now, and it really will help get rid of many of the anklebiters in the domain-name-kiting business.
Once that happens, if enough of them go away, ICANN'll probably just stop there, or maybe they'll wait another 6-12 months, declare it to be a successful test or a Really Good Start, and then finish the job by getting rid of the policy entirely.
No. You're mixing up the required revenue for an individual name to be valuable, the average revenue for the practice of tasting to be profitable with a $0.20 ICANN fee, and the average revenue required with a full no-refund policy on both the ICANN and Registry fees.
An individual name is profitable if it gets more than ~6.42/year, because that's the cost of registering and keeping it, including the ICANN fee and Registrar fee. Tasting as a practice is profitable today if the additional fees charged by a registrar are cheap enough that revenue on profitable names makes enough money to cover those fees, and enough registrars are in a sleazy race-to-the-bottom on prices that you can find some registrar willing to put up with you kiting a lot of names in return for some appropriate share of the profit.
With the 20-cent ICANN fee becoming non-refundable, anybody who's kiting names needs to make enough revenue on the average name they keep to pay for the 20-cent fees on the names they don't keep, as well as on the name itself. Thus, the 6.42 + 199*0.20 = $46.22, assuming the 1-in-200 keep rate. But that's still an average - you'd still keep the names that make more than their $6.42 cost, but if you don't make enough money on average, you'll have to get out of the business (yay!)
This only turns into a $1284 average revenue requirement if you have to pay the entire $6.42 fee on every name you use (which is the cost of buying 200 names - and even that's bogus, because some of those 200 names were names that made non-zero money, but less than $6.42 so they weren't keepers, and in this case you've got no incentive to return them even if they didn't work out.)
The 20-cent fee does kill off one part of the domain-kiting market entirely, though, which is the people who keep rolling over their 5-day registrations and never buy the name. To keep the name for a year, you need to roll it over at least 73 times, and probably more to make sure you don't miss, which will cost at least $14.60, which is more than the $6.42 for registering it legitimately. So even if the whole sleazy process is profitable, at least this fee will cut down on some of the name-kiting transactions that go on.
In fact it's a hard problem if you want 100% coverage. Scotty used to own a large block of space, either through OptInRealBig or one of his other corporate sock puppets, and after that got killed off by anti-spam lawsuits, it took a while before anybody who had that space again was really safe in using it, at least for email.
On the other hand, most people get their IP blocking information from a few large spam-blocker lists, and if those lists can be convinced to remove the block, and the DNS entries for the spammers get cleared out, and if the dozen or so big email services can be convinced to remove it, then it's at least mostly reliable.
As much as I dislike Scotty Richter and his tactics, you can't say he isn't a clever bastard.
The rules for managing pre-ARIN space aren't totally clear, but nobody's worried about them too much because they were mostly owned by large reputable organizations, such as universities and government contractors. (Some of them may need to set the Evil Bit on their packets, but none of them needed to set the Stupid Bit.) In many cases, they've given most of their space back to IANA or ARIN - several universities have returned their Class A/8 space in return for smaller allocations. Also, IANA predates ARIN - while I've got real problems with ICANN's appropriation of Jon Postel's Ghost, and they've delegated most of the policy-making to ARIN, RIPE, APNIC, etc., they're still somewhat in charge.
But there have been a few early-adopters that are no longer in business - and in some cases their IP address space was worth more than their remaining furniture and intellectual property. Does the space revert to IANA if the organization is gone? Probably, but if you can pretend the organization is Not Dead Yet, you might get away with keeping their space. In some cases, you can do that more legitimately than in other cases. (A friend of a friend was the former sysadmin from a defunct early-adopter company that had had a Class B/16 address block, which by the mid-Internet-boom was probably worth $100K. Unfortunately, his ownership of it was dubious enough that he never felt that he could legitimately sell it, and unlike Scotty's newly acquired block of space, it didn't have a corporate shell wrapped around it that he could sell either.)
OptInRealBig and their corporate-shell sock puppets have owned large IP spaces before. It's been a while, so I may have details wrong; if I remember correctly, one of the sock puppets was a "web hosting" company, with lots of "customers", and if one of those "customers" got caught spamming, then they'd get spanked for violating the AUP ("Bad! Bad customer!") - and there was enough IP space that they could keep playing this game for a long time.
One of the devices they mentioned was 400 MHz, one was 266. They're definitely servers, at least in the sense that they have no video (and unlike audio, which you can add with USB, there's no obvious way to add it.)
So it's basically something you'd use for a small home web server, or applications like DNS.
Another alternative is to take an old laptop and add a bigger disk.
I last used Lotus about 10 years ago. I really *wanted* to like it, having used PLATO Notesfiles back in the 1970s (:-), but it was really too big and clumsy for the laptops we used at work, and it spent too much time trying to be a "friendly" GUI to be an actually-useful GUI, especially on smaller screens. It was too bad, especially since MS Mail was still the third-or-fourth worst mail system I'd ever used (IBM PROFS was awful, the early 24x40-screen 300-baud Prodigy wasn't too hot, and we'd once had a Kermit-based homebrew thing that crashed a lot.) Really none of them were as user-friendly as/bin/mail.
Outlook has gotten more bloated, but it really does work much much better than it used to, as long as you've got enough resources to work around its warts. The monolithic-PST-file structure means it's sometimes slow, very hard to back up, and a mess to fix it it gets corrupted, but it doesn't get corrupted very often any more. For server-based mail systems, the bloat means that you need a *lot* of very expensive fast disk farm to store email on, and most corporate IT departments never want to provide enough of that; one reason that Outlook PC storage has become tolerable is that disks have been outrunning Moore's Law for enough years that they're simply Big Enough that it's ok if my current year's PST file is over 2GB.
Outlook has also started to do some really cool things with presence servers, and the server may end up replacing PBXs as we know them, especially because they're doing SIP (to the extent that SIP is a usefully-open protocol.) Their servers are pricy and large, but that's partly because they keep adding more and more functions, and they've certainly seized enough of the calendar market that it's hard to get people to give them up, in spite of lighter and better competition.
David Gale was a Berkeley professor of math, economics, and operations research. He died in March, and there was a memorial service for him today at the college. At least one of the speakers talked about David believing that mathematics was discovered rather than invented, and lots of them talked about how much pleasure he got out of finding interesting problems and elegant solutions to them.
I must confess to not remembering much of the class I took from him 30 years ago (:-), since it's not material I ended up using much over the years, but it did contribute a lot to my mathematical maturity and appreciation of elegance, and had I done more research rather than engineering the matroid theory would have been an especially useful place to work from.
Cable modem service is usually a monopoly in the US, unless you're somewhere that RCN built a second infrastructure - and that's largely because the local governments "helped" people get cable TV by granting franchises to cable providers. Those of us who watched that business evolved generally saw contracts award not to the providers who were most visionary about telecommunications, but usually because of whose brother-in-law got the installation or street paving contracts. The major competitor to cable TV hasn't been another data service (though that's changing with telco fiber-to-the-home and fiber-to-the-block systems) - it's satellite TV, which is fine for TV broadcasts but less than ideal for interactive data. The "cable openness" arguments of the 90s got their technology wrong because they had the wrong policy goals; most of the openness that eventually happened is correct technology, with routing from the head end on up and ISP-friendly wholesale billing contracts.
DSL's different, with far more players at different parts of the OSI protocol stack:
- The dry copper lines are still owned by the monopoly telco,
- the Layer 2 ATM layer may be provided by telco DSLAMs, or may be provided by DSLAMs owned by carriers such as Covad,
- Layer 2 concentration from the DSLAM to the router may be provided by the DSLAM provider or the Layer 3 service provider,
- and the Layer 3 Internet service (including routing, upstream feeds, and extra services such as email and web hosting) may be provided by the DSLAM provider, or may be provided by a third-party ISP such as Speakeasy or Sonic, and the
- Layer 8 services (financial:-) may be provided by the ISP or by some aggregator such as Megapath that sells bundles of service from multiple providers.
(There are a few variants on this such as using frame instead of ATM for Layer 2, or using PPPoE to make it easier to handle billing by over-complicating technology, but most of them don't change the fundamental differences in which provider sets which kinds of policies.)
Traffic throttling and similar harassment isn't primarily a Layer 2 ATM feature - it's an ISP feature usually handled at Layer 3 or 4 of the OSI stack, though occasionally it's done at higher layers (e.g. some detection mechanism and either TCP RSTs or various ICMP distractions.) The amount of oversubscription that happens at Layer 2 may be handled by the DSLAM provider or the router provider, but that's normally done by assigning some committed bandwidth level to each subscriber and doing some kind of fair-queuing with any excess bandwidth, so it doesn't know or care whether you're running one fat FTP session or videoconference or a Layer-3-fairness-evading bundle of many TCP sessions from Bittorrent.
So if you don't like your DSL provider's policies, whether they're P2P throttling or charging too much for static IP addresses or not allowing you to run servers at home or restricting who you can share your wireless router with or whatever, you can get another ISP that'll provide you better services. You may not get more bandwidth than the vanilla telco DSL services, depending on whether you live in a market with multiple DSLAM providers or how good your copper wires to the telco are, and you won't get the latest loss-leader price, but at least in most of the RBOC-served parts of the US, you can certainly get better policies if you want them. My copper's not great (I think I'm still only getting about 1.5/384), but when I decided I wanted a static IP address, I went with sonic.net, whose acceptable use policies were (approximately) "We're selling you *internet* access, not couch-potato service, so do whatever you'd like except for spamming", plus they were doing some cool wireless things, and I could have gotten similarly open policies from Speakeasy and surprisingly reasonable ones from Earthlink. Sonic's a small provider, so they also have the advantage that if I send them email or call them on the phone, I get a response from an actual human with a clue (though sometimes the clue is "looks like a wiring problem, we'll have your telco check it out", and even then, they get to be the ones hassling the telco to get it fixed.)
Yes, there are cans of pressurized dessert topping that are more similar to floor wax than they are to whipped cream. There are other cans of whipped cream that are just fine, typically containing cream, sugar, nitrous oxide, vanilla (usually fake), and often some starches or seaweed polymers. Some are made with heavy cream, some are made with light cream and more polymers (and maybe some powdered dry milk as well.)
The real-cream ones are just fine - they're a bit lighter than the stuff I make at home with a pressure-can and whippets, and of course they're both lighter than the more traditional schlag my mom makes with a Mixmaster.
Cream from the grocery store can also be real or adulterated - if I go to Safeway, the ingredients in the "whipping cream" include the carageenan and various glycerides and maybe some preservatives, and the "fancier" "whipping cream" also has sweetener (I forget if it's sugar or (even worse) corn syrup) and probably vanillan. If I go to Whole Foods or Trader Joe's, the ingredients in the "whipping cream" are "cream", and at most it's ultrapasteurized.
Many years ago, my wife decided to have a board games party for her birthday. Most of our friends are various kinds of techies, including the crypto geek crowd I was hanging out with, and that worked reasonably well. For the invitation, she got a bunch of blank jigsaw puzzles and some plastic silk-screening templates, and silk-screened the invitation onto the puzzles. So the invitations were physically encrypted (:-), and you had to assemble them to get details.
Another memorable techie party was one that my friend Eric threw at Doug's new apartment. Doug had just arrived in town, and most of his furniture hadn't yet. Eric put large rolls of paper on the walls, and started tossing magic markers around for people to write things with. Much entertaining presentation happened, but then there was a Japanese TV crew at the party pointing cameras at people. And the quiet old guy lighting a cigarette on the stove was Timothy Leary (Tim wasn't in great shape by then, but he was still getting around.)
Polio was still around when I was growing up - I'm very glad to be vaccinated, thank you. On the other hand, I got my measles, mumps, and chickenpox immunity the old-fashioned way, and lived through them (though measles in particular can be fatal.) Once they figure out what age you're supposed to get the shingles vaccine at, I'll take it too (it's related to chickenpox.) I've had the pneumonia vaccine, and I'll take it again at whatever age is appropriate.
It's really important that everybody else's kids get vaccinated for the major communicable diseases, because that radically reduces the risk of your kids getting those diseases. But once that's happened, you've got the moral and practical dilemma of how risky it is to your kids and other people's kids to skip vaccination - it's usually not too risky, unless lots of other people do it, at which point it's risky again. Measles seems to reach that stage occasionally.
There are a couple of special cases - tetanus and flu. Tetanus isn't something where vaccinating other kids will protect yours; it's driven by how often kids run into rusty nails and other puncture wounds. And of course flu vaccines are a guess every year - but having the flu sucks enough that I usually get the vaccine.
If you look at a map, there are three reasonable routes to connect Australia to civilization. Either you connect the western side of Aus up to Singapore, or you connect Sydney to sites north and east, and since Indonesia's in the way, if you want to go north, you go toward Guam; if you want to go east, you either stop in Hawaii or keep going to North America. Guam and Hawaii are both starting to get reasonable concentrations of cables stopping there, and Guam provides closer access to Asia than Hawaii does. The Singapore routes are pretty well covered, but it's a really long haul taking that approach to North America, especially if you're starting in Sydney or Melbourne.
Most gasses have boiling points higher than nitrogen's, but there's at least one option between cheap liquid nitrogen and expensive liquid helium, which is liquid Neon, which boils at 24.5 kelvin. The Wikipedia article says it's not cheap, but not as expensive as liquid helium, has better refrigeration properties, and is extracted from air rather than rare sources that risk exhaustion.
Unpredictability doesn't necessarily imply free will, but predictability does imply non-free-will. As you say, randomness doesn't necessarily imply free will either, but at least it leaves you some wiggle room. And Quantum Mechanics is the most convenient current source of randomness around, so dualists, whether they're serious philosophers or kooks, tend to like it.
Unfortunately, the common syllogism works like this
Free will requires something non-predictable and mysterious.
Quantum mechanics is non-predictable and mysterious.
Therefore, Quantum Mechanics gives us free will.
And if you're a New Ager, Quantum Mechanics also is how Consciousness lets us create the universe we want to see if we recite all our affirmations every morning and do stuff with our Chakras!
There are some serious researchers who are looking into how quantum processes affect the brain. The brain's pretty large for quantum effects to show up, but it's still within range, and there are smaller features or processes that are much more within range of quantum effects - but that doesn't mean there's a mechanism to do anything useful with it, and what effects it might have is still a research topic.
There's a story that says that Socrates's servant stole something, and tried to excuse himself by saying that he was fated to steal it and therefore shouldn't be responsible. Socrates's response was that he was fated to beat him, so stop trying to weasel out.
It's not very satisfying philosophically, but if free will doesn't exist, we don't have much choice about acting as if it were true.
I was recently at a Conference on Science and Consciousness. (There are two conferences with roughly the same name; this is the neurologists-and-academics conference, not the Deepak Chopra one.) There was a morning's worth of sessions looking at followups on Libet's work. Most explanations have been been about following neural signals from one brain region to the next, trying to figure out which areas deal with conscious awareness, and how do different signals interfere with each other. (On the other hand, some researchers were looking into the possibility that causality can extend into the past by 100-200ms:-)
ISPs sell to three main markets - content providers, transit users, and eyeballs.
The BBC connects to one or more ISPs - either they pay a content-provider-market or transit-market ISPs, or some eyeball-market ISPs accept their traffic for free because that's cheaper than getting it from other ISPs. The consumer is paying an eyeball-market ISP to get them content, and that ISP either peers with a content-provider-market ISP for free or pays a transit-provider ISP or maybe peers with a transit-provider if they're balanced enough.
I did a traceroute to BBC, and it connected through Telia, a mainly-European ISP. I don't know how many other ISPs BBC is using, or how Tiscali connects to Telia, but Telia seems to be in some combination of eyeball and transit market. (Telia just recovered from a peering squabble with Cogent, a mainly-US content-provider-market ISP.)
If Tiscali's complaining about the impact on their traffic capacity, either they need to work with BBC on how to cache traffic in their network (so they only have to receive it once), or else they need to look into getting a more cost-effective connection to the BBC. Or they could work with BBC on using a P2P distribution system such as Bittorrent.
Works just fine running XP, Outlook, Powerpoint, Word, Excel, Visio, and *one* of the two different browsers (IE), though of course the other browser (Moz2.0.0.12) is the one I do most of my browsing on.... Looks like I should try installing Mozilla 3.0b5.
I agree with the parent article - my work laptop only has 512MB, and it's only been recently that memory's cheap enough for me to bother paying my own cash to upgrade it with DDR1 just to support one over-bloated application. Ok, that one over-bloated application _is_ my browser, and I use it even more than email, and it's not like I'm going to switch to IE for regular use unless I'm really desperate, but Firefox 2.0 is usually using far more RAM on windows I've already closed than Microsoft Office is currently using for the 17 Powerpoint windows, 17 Word documents, 2 Excel spreadsheets, a Visio diagram, and 4 IE windows that I've currently got open now. (Those add up to 13 MB, Outlook is bloating along at 49, and Firefox is currently at 55 only because I just killed the previous 167MB that was swapping to death and reopened 4 tabs of Slashdot. But Firefox is normally well over 100 except when I've just restarted it, and usually turns my machine into a total dog somewhere around 200-300 MB.)
No, it wouldn't kill me to see a few ads. But at least under 2.0, it tends to kill my browser performance - one thing I typically do is open tabs for all the interesting news articles in fark.com, and my machine dogs out badly enough if I do have the ads turned off, and is worse with them on.
Back when ads were just non-blinky banners, they were only a problem because my 14.4kb modem took time to download them, but these days they're singing dancing Flash/Javascript/Popup-trying/Popunder-trying/whatever things that get really annoying, in addition to the privacy implications.
Slashdot Mirror
The cost of those 199 dud attempts is currently near-zero (so it's worthwhile kiting names even if you don't put much effort into predicting their quality very well, because you'll find out soon enough which ones work.) (Techically the price is slightly above zero, because of the cost of money on $6.42 for five days, so it's around $0.01 per name.) So the name-kiters can keep rolling over names today, either buying any name that makes over $6 or continuing to kite names for free while hoping that other DNS-abusers don't steal them at the end of the grace period (which they'd do for names that make
The proposed ICANN fee raises that to $0.20 each, which is where the $46.22 figure comes in, basically $6 for the name you keep and $40 for the duds. That one seems reasonable - you'd still keep any name that makes more than $6.42/year in revenue, but you need to *average* over $46.22 per good name if you're going to stay in business because the duds cost you money. Also, the 20-cent fee for names turns over >73 times/year, which is ~$14.60, so you can't do the continuous-kiting game any more - if a name's not making $6/year, you drop it, and that'll probably cut down a lot on the transaction rates. (Again, that's not really precise, because even some of the duds will make non-zero money - $6/year is about 9 cents per 5-day-grace-period, but you might get some names making 5 cents during the period even if they're not making enough to actually buy.)
But the $1284/year = 200 * $6.42 is different - that's the cost of actually buying all 200 names, as opposed to buying the one good one and only paying the ICANN tasting fee on the duds. It's an even fuzzier figure - it really means that domain-tasting gamblers need to average $6.42/name to stay in the tasting business (which we're guessing won't happen), but again, some of those names might make $3 even if they're not making $6.42, so the best name doesn't need to make the whole $1284.
Maybe you haven't been watching the domain name policies evolve, and assume that they way they are today is both good and necessary. I've been paying attention to them since the mid-80s, and back then it wasn't clear that the market would even accept a system that had a single Root in charge of who could use what names - after all, UUCP worked quite well without it, and the Plan 9 folks had a bunch of really strong arguments for why local-centric naming was better. And I watched the IETF Ad-Hoc Committee work on policies for how to expand the TLD space while the Trademark Gods tried to prevent that from happening (and the bad guys won...) Those "obligations" you're referring to are rules ICANN made, and they'd have been controversial at the time except that ICANN had a policy of ignoring anybody who disagreed with them so controversy wasn't permitted.
The purpose of whois contact information was originally so that if something wasn't working, you could contact the owners of a name to fix it, and once they started charging money for domain names, it was also so that the registrar could reach a name registrant to tell them to pay their bill. That's all - those functions don't require that you provide your True Name, ICBM address, retina print from remaining eyeball, mother's maiden name, first born child's DNA, etc., and they can take care of the billing issues by accepting payment in advance or billing you by email.
From the standpoint of the Trademark Gods, which is what ICANN requires all their registrars to collaborate with, a DNS registration needs to have enough information to get you a subpoena with your name on it in case of a trademark lawsuit (which is where most of the requirement for True Name, Jurisdiction, and Postal Address came from). Various flavors of cops, spammers, and other anti-privacy types think this is just a fine idea, and have tried to extend it. On the other hand, some of the registrars have been providing some private registration services, in some case with some jurisdictional diversification. If you want to spend ~$100/year, you can also get your privacy the old-fashioned way - I've tracked at least one spammer's whois registration to the street address of "The Company Corporation", who've been the canonical business setting up cheap Delaware corporations for over a century, so even if I had successfully sued the spammers for big bucks, their only real assets were a file-folder in a drawer at The Company Corporation, and they could get back in business by spending another $100 for another corporate shell.
Obviously you can't do the same level of dispute resolution or problem-solving if you don't have somewhat usable contact information. It's really helpful to have the technical contact information for a domain be something other than an email address at that domain, because if the DNS is broken, you can't send them the email, but example-dot-com-technical-contact@yahoo.com is perfectly reasonable, and I'd rather not put my home or cellular phone number on a whois page that somebody could call at any time day or night (as opposed to an office phone, which would be perfectly reasonable for a domain owned by a business.) But if there's a legitimate domain name ownership dispute, and the registrar can't reach you by any of the contact mechanisms you choose to provide to them, it's reasonable for the registrar to resolve the dispute by saying you lose, and making it your problem to get the domain name back. Even that's got problems, especially for a personal domain - if you're on vacation for a month visiting your family in India, you might have trouble getting your email, but you also aren't going to see that paper subpoena that the alleged trademark owner put in your home mailbox.
And of *course* you were asking for ICANN to police people
If you want to call your domain I-Got-This-Name-First.com, and somebody else already has that name trademarked, ICANN's policies should prevent you from getting it second, though arguably they could leave room in the DNS structure for disambiguating legitimate multiple uses of names. But if you want to call your website Pirated-Microsoft-Software.com, that's an issue between you and Microsoft to resolve the trademark conflict, and it's certainly not ICANN's business to decide if you really _do_ sell pirated Microsoft software.
Remember that judge who let those Cayman bankers get a temporary injunction against Wikileaks.org's domain name? Even he realized that it was a mistake, in spite of the number of people who don't like Wikileaks posting things that they don't like.
Back when somebody at ICANN invented the Annoying Grace Period that facilitates Domain Tasting, they mandated that the registry and registrars give the money back if somebody returns a domain name within 5 days. I guess it "seemed like a good idea at the time", but of course it's become obvious since then that that was a bad idea.
Think of the 20-cent fee as the beginning of ICANN saying "Yes, we made a mistake" - there's too much ego involved, and too many of the current crop of registrars making a profit on this scammery that they don't want to let go of, for ICANN to simply fix their mistake in one step, because now they've got to compromise with all that political pressure. So they're imposing the policy with their part of the fee now, and it really will help get rid of many of the anklebiters in the domain-name-kiting business.
Once that happens, if enough of them go away, ICANN'll probably just stop there, or maybe they'll wait another 6-12 months, declare it to be a successful test or a Really Good Start, and then finish the job by getting rid of the policy entirely.
An individual name is profitable if it gets more than ~6.42/year, because that's the cost of registering and keeping it, including the ICANN fee and Registrar fee. Tasting as a practice is profitable today if the additional fees charged by a registrar are cheap enough that revenue on profitable names makes enough money to cover those fees, and enough registrars are in a sleazy race-to-the-bottom on prices that you can find some registrar willing to put up with you kiting a lot of names in return for some appropriate share of the profit.
With the 20-cent ICANN fee becoming non-refundable, anybody who's kiting names needs to make enough revenue on the average name they keep to pay for the 20-cent fees on the names they don't keep, as well as on the name itself. Thus, the 6.42 + 199*0.20 = $46.22, assuming the 1-in-200 keep rate. But that's still an average - you'd still keep the names that make more than their $6.42 cost, but if you don't make enough money on average, you'll have to get out of the business (yay!)
This only turns into a $1284 average revenue requirement if you have to pay the entire $6.42 fee on every name you use (which is the cost of buying 200 names - and even that's bogus, because some of those 200 names were names that made non-zero money, but less than $6.42 so they weren't keepers, and in this case you've got no incentive to return them even if they didn't work out.)
The 20-cent fee does kill off one part of the domain-kiting market entirely, though, which is the people who keep rolling over their 5-day registrations and never buy the name. To keep the name for a year, you need to roll it over at least 73 times, and probably more to make sure you don't miss, which will cost at least $14.60, which is more than the $6.42 for registering it legitimately. So even if the whole sleazy process is profitable, at least this fee will cut down on some of the name-kiting transactions that go on.
On the other hand, most people get their IP blocking information from a few large spam-blocker lists, and if those lists can be convinced to remove the block, and the DNS entries for the spammers get cleared out, and if the dozen or so big email services can be convinced to remove it, then it's at least mostly reliable.
The rules for managing pre-ARIN space aren't totally clear, but nobody's worried about them too much because they were mostly owned by large reputable organizations, such as universities and government contractors. (Some of them may need to set the Evil Bit on their packets, but none of them needed to set the Stupid Bit.) In many cases, they've given most of their space back to IANA or ARIN - several universities have returned their Class A
But there have been a few early-adopters that are no longer in business - and in some cases their IP address space was worth more than their remaining furniture and intellectual property. Does the space revert to IANA if the organization is gone? Probably, but if you can pretend the organization is Not Dead Yet, you might get away with keeping their space. In some cases, you can do that more legitimately than in other cases. (A friend of a friend was the former sysadmin from a defunct early-adopter company that had had a Class B
OptInRealBig and their corporate-shell sock puppets have owned large IP spaces before. It's been a while, so I may have details wrong; if I remember correctly, one of the sock puppets was a "web hosting" company, with lots of "customers", and if one of those "customers" got caught spamming, then they'd get spanked for violating the AUP ("Bad! Bad customer!") - and there was enough IP space that they could keep playing this game for a long time.
So it's basically something you'd use for a small home web server, or applications like DNS.
Another alternative is to take an old laptop and add a bigger disk.
Outlook has gotten more bloated, but it really does work much much better than it used to, as long as you've got enough resources to work around its warts. The monolithic-PST-file structure means it's sometimes slow, very hard to back up, and a mess to fix it it gets corrupted, but it doesn't get corrupted very often any more. For server-based mail systems, the bloat means that you need a *lot* of very expensive fast disk farm to store email on, and most corporate IT departments never want to provide enough of that; one reason that Outlook PC storage has become tolerable is that disks have been outrunning Moore's Law for enough years that they're simply Big Enough that it's ok if my current year's PST file is over 2GB.
Outlook has also started to do some really cool things with presence servers, and the server may end up replacing PBXs as we know them, especially because they're doing SIP (to the extent that SIP is a usefully-open protocol.) Their servers are pricy and large, but that's partly because they keep adding more and more functions, and they've certainly seized enough of the calendar market that it's hard to get people to give them up, in spite of lighter and better competition.
I must confess to not remembering much of the class I took from him 30 years ago (:-), since it's not material I ended up using much over the years, but it did contribute a lot to my mathematical maturity and appreciation of elegance, and had I done more research rather than engineering the matroid theory would have been an especially useful place to work from.
18th century mathematician Leopold Kronecker gets credit for saying it.
DSL's different, with far more players at different parts of the OSI protocol stack:
- - The dry copper lines are still owned by the monopoly telco,
- - the Layer 2 ATM layer may be provided by telco DSLAMs, or may be provided by DSLAMs owned by carriers such as Covad,
- - Layer 2 concentration from the DSLAM to the router may be provided by the DSLAM provider or the Layer 3 service provider,
- - and the Layer 3 Internet service (including routing, upstream feeds, and extra services such as email and web hosting) may be provided by the DSLAM provider, or may be provided by a third-party ISP such as Speakeasy or Sonic, and the
- - Layer 8 services (financial
:-) may be provided by the ISP or by some aggregator such as Megapath that sells bundles of service from multiple providers.
(There are a few variants on this such as using frame instead of ATM for Layer 2, or using PPPoE to make it easier to handle billing by over-complicating technology, but most of them don't change the fundamental differences in which provider sets which kinds of policies.)Traffic throttling and similar harassment isn't primarily a Layer 2 ATM feature - it's an ISP feature usually handled at Layer 3 or 4 of the OSI stack, though occasionally it's done at higher layers (e.g. some detection mechanism and either TCP RSTs or various ICMP distractions.) The amount of oversubscription that happens at Layer 2 may be handled by the DSLAM provider or the router provider, but that's normally done by assigning some committed bandwidth level to each subscriber and doing some kind of fair-queuing with any excess bandwidth, so it doesn't know or care whether you're running one fat FTP session or videoconference or a Layer-3-fairness-evading bundle of many TCP sessions from Bittorrent.
So if you don't like your DSL provider's policies, whether they're P2P throttling or charging too much for static IP addresses or not allowing you to run servers at home or restricting who you can share your wireless router with or whatever, you can get another ISP that'll provide you better services. You may not get more bandwidth than the vanilla telco DSL services, depending on whether you live in a market with multiple DSLAM providers or how good your copper wires to the telco are, and you won't get the latest loss-leader price, but at least in most of the RBOC-served parts of the US, you can certainly get better policies if you want them. My copper's not great (I think I'm still only getting about 1.5/384), but when I decided I wanted a static IP address, I went with sonic.net, whose acceptable use policies were (approximately) "We're selling you *internet* access, not couch-potato service, so do whatever you'd like except for spamming", plus they were doing some cool wireless things, and I could have gotten similarly open policies from Speakeasy and surprisingly reasonable ones from Earthlink. Sonic's a small provider, so they also have the advantage that if I send them email or call them on the phone, I get a response from an actual human with a clue (though sometimes the clue is "looks like a wiring problem, we'll have your telco check it out", and even then, they get to be the ones hassling the telco to get it fixed.)
The real-cream ones are just fine - they're a bit lighter than the stuff I make at home with a pressure-can and whippets, and of course they're both lighter than the more traditional schlag my mom makes with a Mixmaster.
Cream from the grocery store can also be real or adulterated - if I go to Safeway, the ingredients in the "whipping cream" include the carageenan and various glycerides and maybe some preservatives, and the "fancier" "whipping cream" also has sweetener (I forget if it's sugar or (even worse) corn syrup) and probably vanillan. If I go to Whole Foods or Trader Joe's, the ingredients in the "whipping cream" are "cream", and at most it's ultrapasteurized.
Many years ago, my wife decided to have a board games party for her birthday. Most of our friends are various kinds of techies, including the crypto geek crowd I was hanging out with, and that worked reasonably well. For the invitation, she got a bunch of blank jigsaw puzzles and some plastic silk-screening templates, and silk-screened the invitation onto the puzzles. So the invitations were physically encrypted (:-), and you had to assemble them to get details.
Another memorable techie party was one that my friend Eric threw at Doug's new apartment. Doug had just arrived in town, and most of his furniture hadn't yet. Eric put large rolls of paper on the walls, and started tossing magic markers around for people to write things with. Much entertaining presentation happened, but then there was a Japanese TV crew at the party pointing cameras at people. And the quiet old guy lighting a cigarette on the stove was Timothy Leary (Tim wasn't in great shape by then, but he was still getting around.)
It's really important that everybody else's kids get vaccinated for the major communicable diseases, because that radically reduces the risk of your kids getting those diseases. But once that's happened, you've got the moral and practical dilemma of how risky it is to your kids and other people's kids to skip vaccination - it's usually not too risky, unless lots of other people do it, at which point it's risky again. Measles seems to reach that stage occasionally.
There are a couple of special cases - tetanus and flu. Tetanus isn't something where vaccinating other kids will protect yours; it's driven by how often kids run into rusty nails and other puncture wounds. And of course flu vaccines are a guess every year - but having the flu sucks enough that I usually get the vaccine.
If you look at a map, there are three reasonable routes to connect Australia to civilization. Either you connect the western side of Aus up to Singapore, or you connect Sydney to sites north and east, and since Indonesia's in the way, if you want to go north, you go toward Guam; if you want to go east, you either stop in Hawaii or keep going to North America. Guam and Hawaii are both starting to get reasonable concentrations of cables stopping there, and Guam provides closer access to Asia than Hawaii does. The Singapore routes are pretty well covered, but it's a really long haul taking that approach to North America, especially if you're starting in Sydney or Melbourne.
Most gasses have boiling points higher than nitrogen's, but there's at least one option between cheap liquid nitrogen and expensive liquid helium, which is liquid Neon, which boils at 24.5 kelvin. The Wikipedia article says it's not cheap, but not as expensive as liquid helium, has better refrigeration properties, and is extracted from air rather than rare sources that risk exhaustion.
Unfortunately, the common syllogism works like this
There are some serious researchers who are looking into how quantum processes affect the brain. The brain's pretty large for quantum effects to show up, but it's still within range, and there are smaller features or processes that are much more within range of quantum effects - but that doesn't mean there's a mechanism to do anything useful with it, and what effects it might have is still a research topic.
It's not very satisfying philosophically, but if free will doesn't exist, we don't have much choice about acting as if it were true.
I was recently at a Conference on Science and Consciousness. (There are two conferences with roughly the same name; this is the neurologists-and-academics conference, not the Deepak Chopra one.) There was a morning's worth of sessions looking at followups on Libet's work. Most explanations have been been about following neural signals from one brain region to the next, trying to figure out which areas deal with conscious awareness, and how do different signals interfere with each other. (On the other hand, some researchers were looking into the possibility that causality can extend into the past by 100-200ms :-)
The BBC connects to one or more ISPs - either they pay a content-provider-market or transit-market ISPs, or some eyeball-market ISPs accept their traffic for free because that's cheaper than getting it from other ISPs. The consumer is paying an eyeball-market ISP to get them content, and that ISP either peers with a content-provider-market ISP for free or pays a transit-provider ISP or maybe peers with a transit-provider if they're balanced enough.
I did a traceroute to BBC, and it connected through Telia, a mainly-European ISP. I don't know how many other ISPs BBC is using, or how Tiscali connects to Telia, but Telia seems to be in some combination of eyeball and transit market. (Telia just recovered from a peering squabble with Cogent, a mainly-US content-provider-market ISP.)
If Tiscali's complaining about the impact on their traffic capacity, either they need to work with BBC on how to cache traffic in their network (so they only have to receive it once), or else they need to look into getting a more cost-effective connection to the BBC. Or they could work with BBC on using a P2P distribution system such as Bittorrent.
Works just fine running XP, Outlook, Powerpoint, Word, Excel, Visio, and *one* of the two different browsers (IE), though of course the other browser (Moz2.0.0.12) is the one I do most of my browsing on.... Looks like I should try installing Mozilla 3.0b5.
If the Feds want to sling bullshit, make sure to answer it appropriately...
I agree with the parent article - my work laptop only has 512MB, and it's only been recently that memory's cheap enough for me to bother paying my own cash to upgrade it with DDR1 just to support one over-bloated application. Ok, that one over-bloated application _is_ my browser, and I use it even more than email, and it's not like I'm going to switch to IE for regular use unless I'm really desperate, but Firefox 2.0 is usually using far more RAM on windows I've already closed than Microsoft Office is currently using for the 17 Powerpoint windows, 17 Word documents, 2 Excel spreadsheets, a Visio diagram, and 4 IE windows that I've currently got open now. (Those add up to 13 MB, Outlook is bloating along at 49, and Firefox is currently at 55 only because I just killed the previous 167MB that was swapping to death and reopened 4 tabs of Slashdot. But Firefox is normally well over 100 except when I've just restarted it, and usually turns my machine into a total dog somewhere around 200-300 MB.)
Back when ads were just non-blinky banners, they were only a problem because my 14.4kb modem took time to download them, but these days they're singing dancing Flash/Javascript/Popup-trying/Popunder-trying/whatever things that get really annoying, in addition to the privacy implications.