ASCII is neither the lowest denominator (that could arguably be a-z) nor the lowest common one (as Chinese does not employ Roman characters at all; neither do Hebrew, Arabic, etc.).
DNS remains available to everyone. DNS does not change. The interpretation of what a certain ASCII-string could also represent changes, but DNS itself does not. We are not reworking DNS here. We are not creating a new root zone, nor are we changing the protocol. This essential protocol remains unchanged.
The.com domain is not affected. The.us domain is not affected. The.de domain is not affected. Second-level domains are, and even those can be represented in ASCII. Accessibility does not suffer any more than it does from a new version of HTML. Not every browser can display it properly, currently. But in the future they will.
ASCII is not universal. A schoolchild in Japan will not learn Roman characters first. They will learn Hiragana and Katakana first, then Kanji, and at some point Romaji and English. Day-to-day life and reading and writing is not done with Roman characters. It's not reasonable to expect those people to adhere to the "lowest common denominator" where that denominator is not actually a real denominator, just something imposed by countries/standard bodies/people who think it is. You can enter ASCII text on those computers, sure (just as you can enter Japanese text on US computers), but everything else you do is not using ASCII.
In any case, this discussion has been had before, and already decided. IDN are already in production in many cctlds. They are not suddenly going to disappear. They are working. "The internet" is not broken./. is still online.:)
You may not be able to spoof paypal.com with the Kanji for water, but you MAY be able to spoof watermoney.co.jp (in Kanji, that is). Phishing doesn't only happen to English-speaking people... Furthermore, the cost/benefit ratio of Anti-Phishing/Unicode-domains is just not one I'd be convinced would go in favour of Anti-Phishing (considering that this would not even solve the most common phishing methods out there -- basically you are proposing a solution to a problem that is not even observed in the marketplace other than by example code).
As for your speed limits analogy -- even that is debatable. Looking at the German Autobahn, for instance, you don't see conclusive proof that its missing speed caps make it significantly more dangerous than highways with a 55mph speedcap (of course speed limits still apply in certain cases like narrow bends, etc.)
In any case though, by "solving" this "problem", one will have to carry around the baggage of that "solution" even long after it is obsolete (when (corporate) identity can be established by better means, or users wake up to the fact that anybody can make any webpage look like anything they like, including any other webpage).
"IDN negates the purpose of DNS completely."
Hyperbole much ? There is a WORLD OF DIFFERENCE between hühnersuppe.de and 22.141.94.114. 99.95% of the visitors of that site will be able to type it in. You basically just said that those 99.95% should use IP addresses since that would be easier (which is bollocks IN ANY CASE, even if 99.95% would not be able to type it in -- DNS provides much more than an isomorphic DomainIP mapping). You also seemed to avoid the point where circumventing missing input methods is not necessary in the usual case, only in special cases.
I think that fleeing dissident MAY have more immediate concerns than discussing the latest Cooking Master Boy episode on that Chinese forum, or even surf the net per se. If you REALLY want to go down that road -- an "economically disadvantaged person" (i.e. poor guy) doesn't necessarily have the means to acquire a computer or internet connection or the time to spend on any social network website. DNS itself is a slap in the face to them, clearly -- since it does not afford for solving social problems. Then again, DNS might stand for Do Not Suffer and just implement social protocols for making that come true. I might be wrong.
You can't map the Kanji for water to an ASCII-lookalike. You can't seriously map the Kanji for waterpond to water, just because it looks somewhat alike. So all you are talking about are some accented characters; not all of which are easily mapped 1:1 as well. How about the ae character ? Does the o with a / through it map to 0 (which is commonly displayed as a O with a / in it), or o ? Or even the phonetic equivalent ? Who decides what character maps to what other character best ? ß maps to b, since it's a beta, kinda ? Or an s since it's the sz character ? or sz even ? ss ? The greeks would have something to say about that... How about n with a ~ ? Sure, it could become a simple n, but really, nj sounds good as well -- after all, we already use two characters to map ae...:P It's impractical and, in my opinion, a shoot first, ask questions later, oops, there is still phishing going on and we have to support this legacy crud for ages kind of action.
Phishing-potential is not solved by mapping everything to ASCII, as I pointed out before. The issue is one of trust, and trust is a HARD beast to solve by technical means (which all anti-phishing tech is trying to do currently, including SSL). I already stated that the argument holds no water with me; You haven't stated anything new to convince me otherwise.
As for "at least they have a chance" -- tough noogies, again. I mean, there is a chance that you can watch videos in a text-only browser as well (if only every text-only browser had libaa linked to it and were capable of decoding h.264). Doesn't mean it's realistic.
It's not stupid to have no ASCII version of a page. A chinese community does not need any ASCII. Their domain does not need any ASCII, either. If you want to access it from the hotel computer in Kansas and anticipate traveling there and doing so, log on to del.icio.us and click on your link, put in your flash drive and use your bookmarks, use google and search for it, etc. There's a chance that it will work. So yeah, you can prepare for that eventuality, JUST like you can prepare for a different power supply. The power adaptor analogy flies:P
As for people having guidance to have an ASCII-only version of their domain available : that's really up to them. If they expect that an ASCII-only domain name could help with their stated goals and attract more traffic, all the more power to them. If they don't care about English-speaking audiences on that Chinese community site, let them... After all, there are plenty of English sites that really don't care about their Chinese viewership and plenty of small-time ISPs that just summarily block any eMail coming from netblocks in China or Korea, since clearly nobody in their right mind would want to exchange email with those loons.:)
Besides, I do not have to accept capabilities as they are. In that case I'd have to accept browsing with Internet Explorer 4 or Netscape 3.51 at Net Cafes. If there's customer demand, the capabilities will change. (and if you really, absolutely, positively, extremely importantly HAVE to communicate with your friends back home in Chinese using Chinese characters without interoperability problems, you could just use your laptop with that shiny new power adapter).
You mischaracterize my argument as being of aesthetic nature. It's not. There are real differences between accented characters and their counterparts, and of course, pictographic scripts that do not map at all to Roman characters. These character are not dangerous. You keep coming back to "but phishers could abuse it !" -- well yeah, and criminals can use caller ID blocking. That's not a good reason to not have that feature.
The "low-level infrastructure" you speak of remains unchanged. Nothing in the bitstream of the DNS protocol changes.
The "internet" is NOT universally usable. You cannot post to Chinese-language discussion boards while they may not understand the English-language ToS on English-speaking boards. It's not universally usable. It never has been, and it never will be -- since everybody uses the 'net differently and for different purposes.
What in the world is the value of an ASCII domain name on an exclusively Japanese website ? (and don't come back to the "I can't type it on the old Windows 98 machine at the Holiday Inn"-defense; That will not be able to display the page either, nor have useful modes of entry for replying to messages or entering search terms.
As for the Kansas hotel net cafe... That's the price you pay when traveling, anywhere. If you go to Europe, you bring power converters, as well, if you fly around the world a lot, you might want to invest in triple-band cellphones, etc. You can prepare for those eventualities before travel. And yeah, in the markets that Japanese-speaking folks are likely to travel, the market will find a solution.
The holy grail of "universal access and accessibility to everything by everyone everywhere at any time" is a myth. ASCII-only TLDs won't make it come true. (and remember, that is what the article is talking about; IDN-second-level-domains are already out there, the bücher.ch example really exists).
We haven't really gone into the political aspects, either. It may sound splendid to the yanks to use ASCII, after all, that's all they ever need. There are ~300 million yanks out there. There are more than a billion Chinese. They might view the whole thing a TAD differently, and just use SHIFT-JIS encoding for everything; would you follow suit if they expected you to do the same on American domains ? After all, it's the logical choice. (I'm playing devil's advocate here, but insisting on ASCII as the one true standard is very, very US-centric. Hell, ASCII doesn't even contain the -Euro symbol. The _inter_net is not an exclusive playpen for the American people -- everybody gets to play, and everybody does. Not everybody speaks English, contrary to popular belief (which will become very apparent if you try to communicate with the NOCs of many ISPs in the Asian region and, to a lesser extent, Europe). Dismissing the non-English scripts as aesthetic bullshit (and that's basically what a flat transformation into something else is saying) will not exactly enamour the international friends. (and before you say that politics should not play a role in these low-level net infrastructure things, just have a look at ICANN and how it is run.)
In any case, phishing needs to be solved in a different manner. ASCII does not prevent it, and you can easily find several Kanji that look very similar as well. There are efforts to figure out how to make "the internet safe", but that'll take a long while, and ignoring local customs will just spell disaster down the road (Korea, China, and Japan could simply just split off their own root DNS systems if the current one does not cater to their needs. A splintered DNS can't be in anybody's best interest -- in that case your horror-scenario will actually come true, you CAN'T access every domain from everywhere anymore, even if you know the IDN-encoding.
I prioritize a multilingual internet over a monolingual culture, yeah. Not everything should conform to the Ascii (note the capitalization:)
The mapping strategy is not supposed to be used by the end-user. Your grandma is not gonna need it. Chances are, your grandm would get the link in the mail, from another websites, or from a google search; grandmas do not usually type in domain names (in fact, if you observe many non-tech-people using the net, very very few will ever type in an URL directly, and many do not really know what the difference between a domain name and a google search is.
Mapping ä, á, à, and â to a may seem fine to you, but it castrates the non-English languages. A Frenchman will not type melangerie when he hears mélangerie on the phone, since melangerie sounds DISTINCTLY different than mélangerie. I'm sure one can find many examples where these accents matter semantically, as well. In fact, let's add ß to the list. das and daß are semantically different words in German. Mapping daß to das would be wrong -- dass would be acceptable, and nobody in their right mind would use dab. Still though, this is not a simple replacement that can be done predictably by a simple algorithm anymore;-)
As for adapting to ascii-only domain names, I'd predict the opposite to become true. IDN domains are live now (the article posted here is about TLDs -- many second level domains have had IDNs for quite a while now. There are plenty of domains using them, and that number is only rising. Another thing that will come into play here is that the importance of domain names declines steadily. It doesn't really matter what your domain is as long as the right keywords in a search engine get people there. Social Bookmarking really does not care at all about the domain name, and several content location descriptor schemes do not rely on domains, either. All this stuff doesn't care about ASCII at all; chances are it does Unicode or even entirely non-roman/latin encodings (there are plenty of sites in China and Korea, for instance, that have not a single word on them that would be legible with just the ASCII character set -- including some of the most popular social sites over there. And yeah, grandma can't get to that discussion about the best fight moves in Tekken, either, since she probably doesn't know how to write the Kanji for Tekken in the search box that is not really labeled search.:)
Of course there are many scripts out there. To make sense of all that, the Unicode guys work hard. Once standardized in Unicode, it's easy to use them everyhwere -- not every machine may dispaly them, but face it, not every machine has to. When you go to amazon.co.jp, you're gonna see a whole bunch of empty character boxes if you don't have Japanese fonts installed, as well. Nobody is advocating for Japanese websites to switch to Romaji just so everybody in the world can read their website. URLs already contain Kanji -- anything after the / can be any unicode character, on any domain. For the common joe there is no discernable reason why that should not apply before the slash, as well -- and indeed it does.
In a perfect world, everybody can access every website from every terminal of every age at all times. That world will not happen and has never happened. If your terminal does not support what you want it to support, you are either gonna change to a terminal that does, or make the owner of that terminal offer the service you want. They'll find that IME installation package, and quickly. There's moolah involved.
I chose göögle.de as a tongue-in-cheek example assuming it would be a search engine. If you can't enter göögle.de or whatever Kanji a Googol translates into.co.jp, chances are you are gonna have trouble entering search terms in that language as well, follow a discussion, or even read the website in question due to missing fonts. göögle.de may be
I know Wingdings ain't in unicode, that's the point. Joe Schmoe doesn't know what Unicode is, let alone that his favourite smileyface isn't in it;-)
The most sensible thing is NOT to map ä to a silent a. The phonetic equivalent is ae, at least in the German language. It might not be in Swedish or other umlaut-using languages. Your fishing-example goes both ways, btw. (and honestly, one should be worried about fishing attempts that abuse multiple mappings for the same character; and whoopdidoo, one is. The paypal.com with the "other" 'a', the whole discussion between the Chinese and Japanese on how to do multilingual domains even though some of their characters are the same but mapped two times, etc. Distinct characters like ä and a should not really pose a problem. In that case one should really forbid using either 1 or l in domain names altogether as well, and that's just stupid. The phishing problem is a separate one IMHO, anyway. People are stupid enough to take paypa1.com or some variaton thereof seriously when it has a SSL c ert -- the root cause of the problem is likely found somewhere else). Phishing is a problem. It's not made easier by IDN, but it's not really made harder, either. So long as everybody and their mother can get an SSL cert that'll not raise a big red flag in every browser for just $15 a year (regardless of who it is registered to), it really doesn't matter. The mark feels safe, the mark enters his password, the mark just lost $1000.
As for the mappings : Have a look at http://en.wikipedia.org/wiki/Internationalized_domain_name ; to make it short, let's use their example : bücher.ch becomes xn-bcher-kvh.ch. If you know how the transliteration/conversion works, you can get there in an internet cafe in vancouver. If you don't, well, tough noogies. You either use one of the gazillion social bookmarking sites, a Google search, your own links on your memory stick, etc.... The problem of an Arabic writer using a US-English computer system is not really what this is all about. There are obviously going to be problems. Guess what -- Japanese people can't actually write Japanese on their webforum when they do not have a way to enter Kanji or Kana on their US-American Keyboards running under a US-American version of Windows withoutt he Japanese IME installed.
In all likelihood, the net-cafe in vancouver won't even have the proper fonts installed to be able to DISPLAY Chinese, Japanese, or Arabic scripts. They are not installed by default in many operating systems, and even if they are, there are bound to be scripts that are not covered.
While you mention a problem, it's not gonna be solved by using ASCII only. The domain name is just part of the puzzle. If I were to want to go to göögle.de and your approach was to be applied, I could now go to the search engine. Tough noogies if I want to search for an Änderungschneiderei, though, or, heavens no, .
And you are only realizing this now ? They've been doing it since the start. I wonder what people who paid for their shiny horse armors felt like after the first shiny 3 minutes of new horse-armor smell.
Yeah, there is. It may be hard to understand for people who only have words that only use ASCII characters (by definition). It's not too much to ask of system administrators to learn English and put up with limitations imposed by monolingual systems. It IS too much to ask this of the common joe. There is no practical reason my grandmother should not be able to go to änderungschneiderei.de, nürnberg.de, überraschungsei.de, etc. -- and neither should people in the Japanese locale be barred from using.co.jp (or, indeed,..) when they want to hear the latest about Tekken.
From a standardization point of view it does not make sense to limit oneself to 127 characters (well, really, 35 + some special characters). It makes sense to standardize how exactly other characters are coded so they work within the current system (which they do -- these "new" domain names are actually encoded in ASCII characters, so if you are a system administrator without a Chinese input mechanism -- not to worry, you can STILL use an identifier you can type with your keyboard if need be.
DNS itsels is not gonna buckle under this. DNS doesn't care whether what it serves up is even human-readable. There is no more load generated by this. If you really really want to get to a domain via an ASCII encoding, you can. It doesn't look as pretty, but hey, it doesn't have to.
(oh, and as for änderungsaufträge vs. anderungsauftrage -- great, you just lost your first customer in Germany -- if I tell you to go to änderungsaufträge on the phone, you would not go to anderungsauftrage. Even if you KNEW that ä is not a proper character in a domain name (and no, the common joe does not know that), you'd probably transliterate it to aenderungsauftraege.de to keep the phonetic footprint the same. Contrary to popular belief, Uber and Über are not pronounced the same, either:) And as already said, äöüß are on the German keyboard. There is nothing to "distinguish" them from characters that would not be on an ASCII keyboard. Nothing whatsoever. Good luck educating people on that fine difference. I'd be surprised if nobody thought of trying to enter Windings as domain name characters before (cute smiley !), let alone such innocuous characters:)
"normal" keyboards is dependent on where you live. What may be "normal" in your region is most definitely not in others. My keyboard contains ääüß without shift or breaking a finger. Yours likely does not (but your backslash does not need you to press Alt Gr, you lucky SOB). Cyrillic keyboards contain mostly non-latin characters in their default setting. Don't get me started on Japanese and Chinese ones.
The point of DNS was never to make it easy for YOU (personally) to type it, but rather to give a descriptive name and IP-independent locator. I mean come on, what's the point of using DNS for a name containing the pronounciation of some of its characters ? Why not just use 31337 IP addresses instead ? 42.42.42.42 -- News that count, stuff that answers ?
I should be able to use änderungsaufträge.de (and I can, the de TLD introduced such URLs a while ago), and people should not break their script to roman/latin just for the TLD when the rest of the URL doesn't use those characters.
You make bad assumptions. Bob and Eve are not the same person. Bob is Eve's computer; Eve is what ordinary people would assume to be Bob (and ordinary people would possibly consider Alice to be Eve, instead).
Your assumption is that Alice thinks the ordinary Bob's computer actually belongs to Bob. According to Alice, it doesn't. Alice owns Bob^WEve's computer, conveniently named Bob; Eve^WAlice wants to talk to Bob, but that silly Bob^Eve who actually owns Eve^WBob (in Bob^WEve's world, not Eve^WAlice's) wants to eavesdrop on the conversation between Alice and her computer Bob which Eve just leased.
Clear now ?:) My braincells are available on eBay, though I think they just dropped in value by virtue of a royal mind-clusterf****.
You recently learned wrong, however. It doesn't matter whether DTAG or Deutsche Post are private enterprises or not; article 10 still applies. Without a law allowing it or a corresponding court order, government officials still cannot go to the Post, open your letters and read them. Article 10 still prevents them from doing that. The German state remains a substantial shareholder in both DTAG and Post, as well. Also, even the private enterprise Deutche Post (and its competitors) are bound by the Postgesetz (as is the DTAG and its competitors by similar laws) which codifies pretty much the same -- i.e. even the private enterprise Post cannot legally read your mail or collect information on your mailing behavior. (See 39 PostG, for instance).
Furthermorse, it's not article 10 that is in question here -- it's the BDSG. We don't need case law for what you describe to still be illegal; the BDSG is pretty clear on what it considers personally identifiable data. While no case law exists, judges are still free to look at other decisions and interpretations.
"I'm pretty confident that 'regular' logging will continue to be alright; the analysis of user behavior is the critical fact here, at least that's how I read it." -- how you read what ? Have you READ the BDSG ? Have a look at http://www.gesetze-im-internet.de/bdsg_1990/index.html... Analysis of user behavior is not a valid reason to log personally identifiable information as per the law. There's an old saying "Wo kein Kläger da kein Richter", though. People don't usually sue Joe Sixpack Webmaster about this stuff.
Some laws proposed recently are silly, I agree. The BDSG is not one of them. The now infamous Hackergesetz, however, is.
A single IP address is not necessarily associated with a single person. Correct. A -> B. This does not imply B->A in any way, shape or form.
The site actually doesn't make that argument, however. It makes the argument that an IP address is not permanently associated with a single person and easily changed for most (most ISPs here assign you a different IP on each login, out of a pool of millions; and most ISPs here do not allow connections to stay connected for longer than 24 hours).
Furthermore, the site states the exact opposite of your assertion a few paragraphs later. IPs are, in fact, personally identifiable to at least the government, police, and intelligence agencies (as well as foreign hostile intelligence agencies and witty hackers of the legal and technical kind) since ISPs store that data (even though they are not required to (yet) and actually currently forbidden to, lawfully.
Last, but not least, your jump from "it's not exactly 1 person == 1 ip" to "it's not personal data at all" is plainly wrong. Take phone numbers as an analogy. You can clearly change phone numbers. Are they suddenly less not associated personally with you, AT ALL ? Take credit card numbers. You can have many of them, or share one with several people, or even change them once they become compromised. Does that make them any less personally identifiable ?
Funny, if only it were true. The law forbids the general collection of all traffic data without due cause. It does not forbid personally identifiable logging on a case-by-case basis to resolve problems and issues.
To sum up... Will somebody PLEASE think of the children ?:P
An argument can be made that IP addresses you use during surfing are data that can be tied to you personally, and as such fall under the strict privacy laws we have here. Don't like it ? Change the law, or work within it. (of course next to nobody actually cares about that law -- the logging you refer to at the ISP level is just as illegal, but has nonetheless been happening for years. DTAG, the biggest German ISP, logs customer IP addresses in violation of the data privacy laws -- they have been sued for it, and at least one customer no longer has his IP logged. Well, only that one customer, that is. Everybody else is still logged.)
You make a valid point about balance. There might still have to be a distinction between government-provided services (such as the site in question) and privately owned enterprises (equal under the law to you and me, while the government is in a clear position of power). In fact, there is; As a citizen you have considerably more power against the government than you do against private enterprise (since the Grundgesetz (our constitution) applies to the disparate legal relationship between the citizenry and the government, not between citizens as such (which corporations are a part of); the most powerful legal tools are reserved for these battles; even if the (strict) privacy laws do not apply, you can still sue based on article 2, for instance (which guarantees the right to personal freedom, an umbrella which includes the right to decide what happens with personally identifiable data about yourself).
You also seem to throw together the laws and practices of the US and Germany into one pot. I'm not certain every phone call ever made is logged here. I'd be surprised if it were, especially when flatrate-type fees are involved.
Personally I prefer tough privacy laws over mandatory data retention and the assumption of general suspiciousness for years. Wiretaps can still be ordered by the court. IMHO, the ISPs and Telcos should not be doing the policework FOR the police. It's not their job.
Web admins can still effectively manage their servers. They may not be able to log IPs for extended periods of time, but they can STILL manage their servers. They can still block IPs. They can still determine the origin of attacks as they are happening. Evidence can still be collected. However, log analysis and data retention for extended periods of time may become shady... Of course you can evade problems in this area by effectively anonymizing the data in a way that cannot be reverted even given auxiliary data -- once that is done, the data is no longer personally identifiable and does no longer fall under the strict privacy laws.
Indeed it should. It is not, however, a security issue, and as such it falls under the system administrator's job to... you know... administrate his or her system.
Nobody is screaming for an update to Slackware 1.0, either. Same thing.
There are many reasons that airports have special significance:
One by one, eh ?
Terrorists have a track record of targeting them
Also, Skyscrapers, Pizza places, night clubs, crowded areas, buses, cars, subways, college campuses, palaces, houses of the wealthy, and no end of other things. You see, "terrorists" are like snowflakes. None are alike. Some batshit crazy terrorist will find something you'd find to be without any special significance significant enough to blow up. Be afraid. Be very afraid.
Come on, that doesn't even come close to justification or counterpoint.
A small bomb in a park can kill a few people. A small bomb on an airplane can kill a few hundred.
A small bomb in a space station can kill a few less people. A bomb in a park while there's a music festival there can kill more people than there are in many planes. A bomb in my backyard can kill a squirrel. A bomb in a bus can kill slightly less people than in a plane, slightly more than in an empty park.
What, exactly, is your point ? What makes the airport special ? Bombs can kill people. We established that. How is killing people at point A somehow more significant at point B ? (and note that you can probably detonate a minivan full of explosives in a park, while you'll probably have less thereof on a plane. If you really want to optimize your terrorist attack for the variable X, where X is number of people killed, and X is to be equal to some set number, you can probably figure out a plan to do just that, use just the right kind and amount of explosives, etc.
If you set off a bomb at a gas station or a grocery store, people will still buy gas and groceries. But most business travel and all tourism is discretionary. So one bomb can result in the loss of millions or even billions of dollars of economic actitivity.
You are elevating fearmongering to an art.
Let's go into a hypthetical. Let's say we set off a bomb at a walmart located close to the White house, and then another 5 minutes later at a gas station on the other side of the white house. Mix in some napalm so that it really burns for a long time and is nicely visible. What kind of impact, do you think, would such a thing have if there happened to be a couple of news networks at the scene within seconds, without much other information ? PANIC ! PANIC ! Washington is under attack ! Let's dig in and only come out when the president says it's OK !
Well placed bombs in locations other than airports can easily disrupt the economy. Hell, meaningful terrorism of any kind can do that. Planes are not special in that respect. What if you blow up a train ? (make it contain some radioactive waste, just for kicks...) What if there's just a regular accident, no terrorism involved ? The same things can happen. It doesn't really make the airport special and holy. The terrorists may have to find a different target, but a target they will find. Those batshit crazy buggers aren't always without intelligence.
None of your arguments actually show why an airport is more significant than a busy bus or train station w.r.t. the amount of force, searches, and general suspicion. Heck, my laptop bag contains some scaaaaaary electronics (i.e. home-built and not necessarily encased). I'm looking forward to explaining how it all works to the hopefully friendly guy pointing a gun at my face, one day.
Let's knock out the obvious ones first, shall we ?
http://www.megaupload.com/ has one offering, 250gbyte. Prepaid for one year it's 50 Euros (or whatever their site says for the US locale). That's 70 bucks. You/can/ use this as a storage-only service, but of course you can also use it for distribution and such -- no transfer limits. Rapidshare.com has similar offerings (with "unlimited" storage but a 5gbyte/day cap) at similar prices. Both of these rely heavily on customers infringing copyrights, so it's anyone's guess how long they'll stay around. Both also use somewhat nonstandard file deposit and file delivery methods. There are countless others in that market (oxedion, mediafire, upfile, rs.de, filefront, etc., all with varying foci).
The regular webhosting market has things like this to offer as well. http://www.dreamhost.com/ : The cheapest plan, at one year prepayment, would be around 120 bucks and offer 145gb of space. I say would since you can use their promo codes (check the forums) to almost triple the space or drop the price to a lot less. So that's 400 gb of storage, a couple terabytes of transfer a month, and some processing power to boot (WebDAV/FTP/SFTP/SCP/rsync/etc. are all possible). I imagine competitors to DH will have similar offerings space-wise. We're looking at around a fourth the price for almost double the storage space. Don't you dare yell "overselling" -- Google does, too.
If you can be bothered with some cumbersome setup (to laypeople, anyway), Amazon S3 will get you storage space for $0.15/gb/month, plus traffic ($0.18/gb). If you actually use 250gb, the price will be comparable to Google for storage alone (i.e. no transfers other than the initial incoming transfer); the difference is that you get charged by the byte, not in large pre-paid packages. If you use 1gb and transfer it twice, you pay $0.51 that month. Also consider that if you use less than the 250gb Google offering, you're probably get away cheaper (since the smaller Google plans are comparatively more expensive while Amazon's offering exhibits a linear price curve over the amount of storage used).
The value Google's space has is probably the integration with its applications -- Picasa, for instance, lacks decent online functionality using standard protocols -- and Google will probably deliver GREAT online functionality with their own service.
If all you really need is a foolproof backup, open up an FTP and let the world mirror it. I wonder who would do such a thing...
Have you priced the official disks? Have you ever used OpenSSH? If so, have you ever given anything back to the creators and maintainers of OpenSSH (OpenBSD)? Beside the point. They choose to do things like they are, and that's their prerogative. I fully support their right to do so. It does not mean that I have to agree with or like it, or even consider "giving back". See further down on why.
This attitude pisses me off. If you were actually using OpenBSD, you'd be willing to fork over a few buck to get the disks. Don't assume. It just makes you look like an idiot.
But you're not using it. The amount of time spent to produce such a high quality OS is worth the money in my book. Precisely. In your book. Don't tell me what to do on my books, or anybody else's books.
The other thing that pisses me off is that OpenBSD doesn't have a millionaire patron. But they do have Sun, Cisco, etc shipping their software (OpenSSH) withouth even bothering to contribute to the foundation. Kinda cheap, huh? Don't like it ? Change the software's license.
No, really. If you don't actually mean "it's free for whatever you want to use it for", don't release it under a license saying that. It's quite simple, really. Bitching and moaning after people take you up on your offer is just childish. Sure it would be nice of Sun to help the project financially. Same goes for Cisco. But you damn well better don't EXPECT them to. If you give a gift to the world, it is exactly that. A gift. No strings attached. If you want strings, freaking spell it out in the license. In that case it's not a gift anymore, but a product for sale with you expecting compensation.
Maybe that's why they charge for their install disks.
You clearly know nothing about OpenBSD. Just because I don't agree with everything they do 100% or wouldn't follow Theo everywhere does not imply my lack of knowledge about or usage of OpenBSD. But hey, it makes for a cheap ad hominem attack, doesn't it...
TrueCrypt, while a fine piece of software, could never offer what this stick APPEARS to offer -- namely data eradication in the case of an attacker attempting a wrong password too many times in a row. This would protect against brute-forcing the password, possibly on a cluster of computers.
TrueCrypt has been broken before, just not in the way you believe. The methods they use are sound, but no amount of secure encryption can protect you from a brute-force attack on weak passphrases. The idea with the secustick is that you don't even get access to the ciphertext without credentials being present (and a truly decent stick would encrypt the data with the given password anyway, making the debugger a useless tool against it; people could possibly still disassemble the stick and try to dump the memory which would give them the ciphertext to attempt brute-force attacks on, but that's another barrier to break -- especially if the stick is built in a way that would hamper efforts to disassemble it without destroying the data.
Ah well, there are enough knuckleheads who will buy anything that says "Secure ! We promise !" without actually understanding whether what they're buying is snake-oil. Capitalism rocks, that way.
Again, with IPTV, there's a realistic limit on how much you're going to watch in a given month. In other words, there is a clear definition of what "unlimited" means in that context. "Unlimited" has no context. The meaning of the word implies no presence of any limit whatsoever. If I sell you an unlimited supply of air, I damn well better be prepared to farm the galaxy for it.
You can change the meaning with adjectives. That has not happened here.
How is it abuse if I have the tube running 24/7 as background noise ? Don't assume that just because YOU can't imagine a way to keep your TV on 24/7 there isn't anybody who might legitimately do so. Hey, maybe they are too lazy to press the off button, or they'd rather mute it instead. That's not abuse.
And let's not forget that the cable company can easily throttle back your general Internet bandwidth in the case that you're using excessive IPTV bandwidth. Unless, say, the IPTV is not delivered by the cable company but by a third party. Whooops. Triple play and all that crap sound nice on paper for the cable company, customers expect free choice of competitors, though.
Now as far as the people in the article, they are CLEARLY using their cellphones as a general Internet connection for their computers. This is FORBIDDEN by the cellphone TOS unless you sign up for a different plan. I just flat out don't believe that someone used 5GB of bandwidth in a month by checking email and surfing web pages using ONLY their cellphone. You know that emails can easily be 100m or more in size ? Webpages can contain media ? "surfing webpages" is not a clearly defined term ?
You also forget that intranet access is perfectly acceptable. My intranet has tons of data to be perused. Note that " downloading legally acquired songs" (term from the ToS) is indeed allowed. Whoopdidoo, FLAC is big these days.
"Anyone using more than 5 GB per line in a given month is presumed to be using the service in a manner prohibited above,"
Is quite interesting as well. It's a blatant lie, all around. They could just as well have written
"Anyone using more than 5 GB per line in a given month is costing us more money than we alotted for this 'unlimited' plan, so we cut them off."
Maybe so, they just kinda, sorta forgot proper IPv6 tunneling options. You know, the stuff you need to get it going on the current net infrastructure. Also the stuff that Windows XP does without a hitch. See
If you want to use tunnel brokers, you'll have to install third-party VPN drivers (OpenVPN). This is a feature that SHOULD have been there from the start if Microsoft cared about full, usable IPv6 integration.
> There may still be people who like Windows 98, but there aren't people who use Windows XP, and say "Gee, I wish I was using 98 instead."
I know a few people like that. Windows 98 still runs better in low-memory situations, does everything "they" need, and quickly at that. 2k/XP are better operating systems overall IMO, but gee, they are not the hammer to every nail:)
> So shall it be with Vista when it matures.
So operating systems are kind of like wine ? What if Vista is a bad year ?
You write : "If you are going to use math to make an argument, you damn well better get it right. Hmm, let's see. A typical T1 line delivers data at the rate of 1536 Kbps (don't bother about the extra 8Kbps, OK?). So, that's 1536000 / 8 / 1024 ^ -2, or a whopping.1831 GBps, or 10.986 GB/min, or 659.16 GB/hr, or 15819.84 GB/day, or 474,595.2 GB/mo."
How you get to 1536000 / 8 / 1024 ^ -2 to 0.1831 GBps, God only knows. Bits or Bytes, 10^-2 or 1024^-2, etc.
KISS. 1536kbit/s = 192kbyte/s. A month has 30*24*60*60 = 2592000 seconds. 192*2592000 = 497664000. So you use 497664000 Kilobytes per month. That's 474 Gigabytes. To arrive at your claimed number, we'd have to use 1000 T1s or 1.5 gigabit/s. That is a tad more expensive, even if we are not talking about the '90s. Also note that in networking, usually we deal with powers of 10 instead of 2.
So it's not.0003 USD, but 0.003 USD. (and even.0003 USD would be wrong if you assumed terabytes instead.) All the rest of your math is pretty much wrong due to these rather grave mistakes.
You also fail to realize that you don't buy gigabytes, but bandwidth. More to the point, the provider and its backbone most certainly don't consider gigabytes per month, but bits/second of bandwith. A 1.5mbit/s line does not mean that you actually transfer 474 gigabytes a month. It just means you have the capability to should you be able to continually fill it. Thinking of this in terms of gigabytes is just easier on the (usually somewhat dense) marketing department, and is a lot easier to convey to stupid prospective customers. Let's face it, how many of em would understand 95th percentile billing for burstable bandwidth, or even why that's not such a bad idea ?
Last, but not least, maybe some more data. You can get 1mbit/s of bandwidth with 95th percentile billing for $40. Some are cheaper, some are more expensive, but that's a good baseline of what you can get in a datacenter. If you cap your bandwidth at 1mbit/s, that 95th percentile is also 1mbit/s, so that's valid. Rethink your calculations now, and take into account that a DC does not have to roll out the last mile, unlike ISPs. ISPs are still usually corporations and want to make money, some are greedy, and many could offer their services cheaper. Obviously if it really was that much cheaper as you claim, go compete with them. I mean, come on, if you can do it 10 times cheaper, just halve their prices and still make 5 times as much as you spend ! So easy, eh ? Thought so.
Asiding the problem that BitTorrent interclient communication is generally not UDP, this would also increase the tracker load manifold. Currently, each BitTorrent client contacts the tracker every 5-60 minutes or so. Using this technique could quite possibly increase this to a few packets per second per client on busy swarms (assuming, for the sake of simplicity, 1 pps per client, that would result in a 300fold to 3600fold increase in pps on the tracker. A tracker of ~50000 peers currently serves 300-350 requests per second. Increasing that by a few orders of magnitude is not exactly high on a tracker operators' agenda (though it might be worth it for corporate content distribution systems).
As a sidenote, SIP implementations (internet telephony) use the STUN protocol to much the same effect.
Slashdot Mirror
ASCII is neither the lowest denominator (that could arguably be a-z) nor the lowest common one (as Chinese does not employ Roman characters at all; neither do Hebrew, Arabic, etc.).
.com domain is not affected. The .us domain is not affected. The .de domain is not affected. Second-level domains are, and even those can be represented in ASCII. Accessibility does not suffer any more than it does from a new version of HTML. Not every browser can display it properly, currently. But in the future they will.
/. is still online. :)
DNS remains available to everyone. DNS does not change. The interpretation of what a certain ASCII-string could also represent changes, but DNS itself does not. We are not reworking DNS here. We are not creating a new root zone, nor are we changing the protocol. This essential protocol remains unchanged.
The
ASCII is not universal. A schoolchild in Japan will not learn Roman characters first. They will learn Hiragana and Katakana first, then Kanji, and at some point Romaji and English. Day-to-day life and reading and writing is not done with Roman characters. It's not reasonable to expect those people to adhere to the "lowest common denominator" where that denominator is not actually a real denominator, just something imposed by countries/standard bodies/people who think it is.
You can enter ASCII text on those computers, sure (just as you can enter Japanese text on US computers), but everything else you do is not using ASCII.
In any case, this discussion has been had before, and already decided. IDN are already in production in many cctlds. They are not suddenly going to disappear. They are working. "The internet" is not broken.
You may not be able to spoof paypal.com with the Kanji for water, but you MAY be able to spoof watermoney.co.jp (in Kanji, that is). Phishing doesn't only happen to English-speaking people ...
Furthermore, the cost/benefit ratio of Anti-Phishing/Unicode-domains is just not one I'd be convinced would go in favour of Anti-Phishing (considering that this would not even solve the most common phishing methods out there -- basically you are proposing a solution to a problem that is not even observed in the marketplace other than by example code).
As for your speed limits analogy -- even that is debatable. Looking at the German Autobahn, for instance, you don't see conclusive proof that its missing speed caps make it significantly more dangerous than highways with a 55mph speedcap (of course speed limits still apply in certain cases like narrow bends, etc.)
In any case though, by "solving" this "problem", one will have to carry around the baggage of that "solution" even long after it is obsolete (when (corporate) identity can be established by better means, or users wake up to the fact that anybody can make any webpage look like anything they like, including any other webpage).
"IDN negates the purpose of DNS completely."
Hyperbole much ?
There is a WORLD OF DIFFERENCE between hühnersuppe.de and 22.141.94.114. 99.95% of the visitors of that site will be able to type it in. You basically just said that those 99.95% should use IP addresses since that would be easier (which is bollocks IN ANY CASE, even if 99.95% would not be able to type it in -- DNS provides much more than an isomorphic DomainIP mapping).
You also seemed to avoid the point where circumventing missing input methods is not necessary in the usual case, only in special cases.
I think that fleeing dissident MAY have more immediate concerns than discussing the latest Cooking Master Boy episode on that Chinese forum, or even surf the net per se.
If you REALLY want to go down that road -- an "economically disadvantaged person" (i.e. poor guy) doesn't necessarily have the means to acquire a computer or internet connection or the time to spend on any social network website. DNS itself is a slap in the face to them, clearly -- since it does not afford for solving social problems. Then again, DNS might stand for Do Not Suffer and just implement social protocols for making that come true. I might be wrong.
I think you are missing the point.
... How about n with a ~ ? Sure, it could become a simple n, but really, nj sounds good as well -- after all, we already use two characters to map ae ... :P
:P
... After all, there are plenty of English sites that really don't care about their Chinese viewership and plenty of small-time ISPs that just summarily block any eMail coming from netblocks in China or Korea, since clearly nobody in their right mind would want to exchange email with those loons. :)
:)
You can't map the Kanji for water to an ASCII-lookalike. You can't seriously map the Kanji for waterpond to water, just because it looks somewhat alike. So all you are talking about are some accented characters; not all of which are easily mapped 1:1 as well. How about the ae character ? Does the o with a / through it map to 0 (which is commonly displayed as a O with a / in it), or o ? Or even the phonetic equivalent ? Who decides what character maps to what other character best ? ß maps to b, since it's a beta, kinda ? Or an s since it's the sz character ? or sz even ? ss ? The greeks would have something to say about that
It's impractical and, in my opinion, a shoot first, ask questions later, oops, there is still phishing going on and we have to support this legacy crud for ages kind of action.
Phishing-potential is not solved by mapping everything to ASCII, as I pointed out before. The issue is one of trust, and trust is a HARD beast to solve by technical means (which all anti-phishing tech is trying to do currently, including SSL). I already stated that the argument holds no water with me; You haven't stated anything new to convince me otherwise.
As for "at least they have a chance" -- tough noogies, again. I mean, there is a chance that you can watch videos in a text-only browser as well (if only every text-only browser had libaa linked to it and were capable of decoding h.264). Doesn't mean it's realistic.
It's not stupid to have no ASCII version of a page. A chinese community does not need any ASCII. Their domain does not need any ASCII, either. If you want to access it from the hotel computer in Kansas and anticipate traveling there and doing so, log on to del.icio.us and click on your link, put in your flash drive and use your bookmarks, use google and search for it, etc. There's a chance that it will work. So yeah, you can prepare for that eventuality, JUST like you can prepare for a different power supply. The power adaptor analogy flies
As for people having guidance to have an ASCII-only version of their domain available : that's really up to them. If they expect that an ASCII-only domain name could help with their stated goals and attract more traffic, all the more power to them. If they don't care about English-speaking audiences on that Chinese community site, let them
Besides, I do not have to accept capabilities as they are. In that case I'd have to accept browsing with Internet Explorer 4 or Netscape 3.51 at Net Cafes. If there's customer demand, the capabilities will change.
(and if you really, absolutely, positively, extremely importantly HAVE to communicate with your friends back home in Chinese using Chinese characters without interoperability problems, you could just use your laptop with that shiny new power adapter).
Ah well
You mischaracterize my argument as being of aesthetic nature. It's not. There are real differences between accented characters and their counterparts, and of course, pictographic scripts that do not map at all to Roman characters.
... That's the price you pay when traveling, anywhere. If you go to Europe, you bring power converters, as well, if you fly around the world a lot, you might want to invest in triple-band cellphones, etc.
These character are not dangerous. You keep coming back to "but phishers could abuse it !" -- well yeah, and criminals can use caller ID blocking. That's not a good reason to not have that feature.
The "low-level infrastructure" you speak of remains unchanged. Nothing in the bitstream of the DNS protocol changes.
The "internet" is NOT universally usable. You cannot post to Chinese-language discussion boards while they may not understand the English-language ToS on English-speaking boards. It's not universally usable. It never has been, and it never will be -- since everybody uses the 'net differently and for different purposes.
What in the world is the value of an ASCII domain name on an exclusively Japanese website ? (and don't come back to the "I can't type it on the old Windows 98 machine at the Holiday Inn"-defense; That will not be able to display the page either, nor have useful modes of entry for replying to messages or entering search terms.
As for the Kansas hotel net cafe
You can prepare for those eventualities before travel. And yeah, in the markets that Japanese-speaking folks are likely to travel, the market will find a solution.
The holy grail of "universal access and accessibility to everything by everyone everywhere at any time" is a myth. ASCII-only TLDs won't make it come true. (and remember, that is what the article is talking about; IDN-second-level-domains are already out there, the bücher.ch example really exists).
We haven't really gone into the political aspects, either. It may sound splendid to the yanks to use ASCII, after all, that's all they ever need. There are ~300 million yanks out there. There are more than a billion Chinese. They might view the whole thing a TAD differently, and just use SHIFT-JIS encoding for everything; would you follow suit if they expected you to do the same on American domains ? After all, it's the logical choice. (I'm playing devil's advocate here, but insisting on ASCII as the one true standard is very, very US-centric. Hell, ASCII doesn't even contain the -Euro symbol. The _inter_net is not an exclusive playpen for the American people -- everybody gets to play, and everybody does. Not everybody speaks English, contrary to popular belief (which will become very apparent if you try to communicate with the NOCs of many ISPs in the Asian region and, to a lesser extent, Europe). Dismissing the non-English scripts as aesthetic bullshit (and that's basically what a flat transformation into something else is saying) will not exactly enamour the international friends.
(and before you say that politics should not play a role in these low-level net infrastructure things, just have a look at ICANN and how it is run.)
In any case, phishing needs to be solved in a different manner. ASCII does not prevent it, and you can easily find several Kanji that look very similar as well. There are efforts to figure out how to make "the internet safe", but that'll take a long while, and ignoring local customs will just spell disaster down the road (Korea, China, and Japan could simply just split off their own root DNS systems if the current one does not cater to their needs. A splintered DNS can't be in anybody's best interest -- in that case your horror-scenario will actually come true, you CAN'T access every domain from everywhere anymore, even if you know the IDN-encoding.
I prioritize a multilingual internet over a monolingual culture, yeah. Not everything should conform to the Ascii (note the capitalization :)
;-)
:)
.co.jp, chances are you are gonna have trouble entering search terms in that language as well, follow a discussion, or even read the website in question due to missing fonts.
The mapping strategy is not supposed to be used by the end-user. Your grandma is not gonna need it. Chances are, your grandm would get the link in the mail, from another websites, or from a google search; grandmas do not usually type in domain names (in fact, if you observe many non-tech-people using the net, very very few will ever type in an URL directly, and many do not really know what the difference between a domain name and a google search is.
Mapping ä, á, à, and â to a may seem fine to you, but it castrates the non-English languages. A Frenchman will not type melangerie when he hears mélangerie on the phone, since melangerie sounds DISTINCTLY different than mélangerie. I'm sure one can find many examples where these accents matter semantically, as well. In fact, let's add ß to the list. das and daß are semantically different words in German. Mapping daß to das would be wrong -- dass would be acceptable, and nobody in their right mind would use dab. Still though, this is not a simple replacement that can be done predictably by a simple algorithm anymore
As for adapting to ascii-only domain names, I'd predict the opposite to become true. IDN domains are live now (the article posted here is about TLDs -- many second level domains have had IDNs for quite a while now. There are plenty of domains using them, and that number is only rising. Another thing that will come into play here is that the importance of domain names declines steadily. It doesn't really matter what your domain is as long as the right keywords in a search engine get people there. Social Bookmarking really does not care at all about the domain name, and several content location descriptor schemes do not rely on domains, either. All this stuff doesn't care about ASCII at all; chances are it does Unicode or even entirely non-roman/latin encodings (there are plenty of sites in China and Korea, for instance, that have not a single word on them that would be legible with just the ASCII character set -- including some of the most popular social sites over there. And yeah, grandma can't get to that discussion about the best fight moves in Tekken, either, since she probably doesn't know how to write the Kanji for Tekken in the search box that is not really labeled search.
Of course there are many scripts out there. To make sense of all that, the Unicode guys work hard. Once standardized in Unicode, it's easy to use them everyhwere -- not every machine may dispaly them, but face it, not every machine has to. When you go to amazon.co.jp, you're gonna see a whole bunch of empty character boxes if you don't have Japanese fonts installed, as well. Nobody is advocating for Japanese websites to switch to Romaji just so everybody in the world can read their website. URLs already contain Kanji -- anything after the / can be any unicode character, on any domain. For the common joe there is no discernable reason why that should not apply before the slash, as well -- and indeed it does.
In a perfect world, everybody can access every website from every terminal of every age at all times. That world will not happen and has never happened. If your terminal does not support what you want it to support, you are either gonna change to a terminal that does, or make the owner of that terminal offer the service you want. They'll find that IME installation package, and quickly. There's moolah involved.
I chose göögle.de as a tongue-in-cheek example assuming it would be a search engine. If you can't enter göögle.de or whatever Kanji a Googol translates into
göögle.de may be
I know Wingdings ain't in unicode, that's the point. Joe Schmoe doesn't know what Unicode is, let alone that his favourite smileyface isn't in it ;-)
... The problem of an Arabic writer using a US-English computer system is not really what this is all about. There are obviously going to be problems. Guess what -- Japanese people can't actually write Japanese on their webforum when they do not have a way to enter Kanji or Kana on their US-American Keyboards running under a US-American version of Windows withoutt he Japanese IME installed.
The most sensible thing is NOT to map ä to a silent a. The phonetic equivalent is ae, at least in the German language. It might not be in Swedish or other umlaut-using languages. Your fishing-example goes both ways, btw.
(and honestly, one should be worried about fishing attempts that abuse multiple mappings for the same character; and whoopdidoo, one is. The paypal.com with the "other" 'a', the whole discussion between the Chinese and Japanese on how to do multilingual domains even though some of their characters are the same but mapped two times, etc. Distinct characters like ä and a should not really pose a problem. In that case one should really forbid using either 1 or l in domain names altogether as well, and that's just stupid. The phishing problem is a separate one IMHO, anyway. People are stupid enough to take paypa1.com or some variaton thereof seriously when it has a SSL c
ert -- the root cause of the problem is likely found somewhere else).
Phishing is a problem. It's not made easier by IDN, but it's not really made harder, either. So long as everybody and their mother can get an SSL cert that'll not raise a big red flag in every browser for just $15 a year (regardless of who it is registered to), it really doesn't matter. The mark feels safe, the mark enters his password, the mark just lost $1000.
As for the mappings : Have a look at http://en.wikipedia.org/wiki/Internationalized_domain_name ; to make it short, let's use their example : bücher.ch becomes xn-bcher-kvh.ch. If you know how the transliteration/conversion works, you can get there in an internet cafe in vancouver. If you don't, well, tough noogies. You either use one of the gazillion social bookmarking sites, a Google search, your own links on your memory stick, etc.
In all likelihood, the net-cafe in vancouver won't even have the proper fonts installed to be able to DISPLAY Chinese, Japanese, or Arabic scripts. They are not installed by default in many operating systems, and even if they are, there are bound to be scripts that are not covered.
While you mention a problem, it's not gonna be solved by using ASCII only. The domain name is just part of the puzzle. If I were to want to go to göögle.de and your approach was to be applied, I could now go to the search engine. Tough noogies if I want to search for an Änderungschneiderei, though, or, heavens no, .
And you are only realizing this now ? They've been doing it since the start. I wonder what people who paid for their shiny horse armors felt like after the first shiny 3 minutes of new horse-armor smell.
Yeah, there is. It may be hard to understand for people who only have words that only use ASCII characters (by definition). It's not too much to ask of system administrators to learn English and put up with limitations imposed by monolingual systems. It IS too much to ask this of the common joe. There is no practical reason my grandmother should not be able to go to änderungschneiderei.de, nürnberg.de, überraschungsei.de, etc. -- and neither should people in the Japanese locale be barred from using .co.jp (or, indeed, ..) when they want to hear the latest about Tekken.
:) And as already said, äöüß are on the German keyboard. There is nothing to "distinguish" them from characters that would not be on an ASCII keyboard. Nothing whatsoever. Good luck educating people on that fine difference. I'd be surprised if nobody thought of trying to enter Windings as domain name characters before (cute smiley !), let alone such innocuous characters :)
From a standardization point of view it does not make sense to limit oneself to 127 characters (well, really, 35 + some special characters). It makes sense to standardize how exactly other characters are coded so they work within the current system (which they do -- these "new" domain names are actually encoded in ASCII characters, so if you are a system administrator without a Chinese input mechanism -- not to worry, you can STILL use an identifier you can type with your keyboard if need be.
DNS itsels is not gonna buckle under this. DNS doesn't care whether what it serves up is even human-readable. There is no more load generated by this. If you really really want to get to a domain via an ASCII encoding, you can. It doesn't look as pretty, but hey, it doesn't have to.
(oh, and as for änderungsaufträge vs. anderungsauftrage -- great, you just lost your first customer in Germany -- if I tell you to go to änderungsaufträge on the phone, you would not go to anderungsauftrage. Even if you KNEW that ä is not a proper character in a domain name (and no, the common joe does not know that), you'd probably transliterate it to aenderungsauftraege.de to keep the phonetic footprint the same. Contrary to popular belief, Uber and Über are not pronounced the same, either
What baffoon modded this insightful ?
"normal" keyboards is dependent on where you live. What may be "normal" in your region is most definitely not in others. My keyboard contains ääüß without shift or breaking a finger. Yours likely does not (but your backslash does not need you to press Alt Gr, you lucky SOB). Cyrillic keyboards contain mostly non-latin characters in their default setting. Don't get me started on Japanese and Chinese ones.
The point of DNS was never to make it easy for YOU (personally) to type it, but rather to give a descriptive name and IP-independent locator. I mean come on, what's the point of using DNS for a name containing the pronounciation of some of its characters ? Why not just use 31337 IP addresses instead ? 42.42.42.42 -- News that count, stuff that answers ?
I should be able to use änderungsaufträge.de (and I can, the de TLD introduced such URLs a while ago), and people should not break their script to roman/latin just for the TLD when the rest of the URL doesn't use those characters.
You make bad assumptions. Bob and Eve are not the same person. Bob is Eve's computer; Eve is what ordinary people would assume to be Bob (and ordinary people would possibly consider Alice to be Eve, instead).
:) My braincells are available on eBay, though I think they just dropped in value by virtue of a royal mind-clusterf****.
Your assumption is that Alice thinks the ordinary Bob's computer actually belongs to Bob. According to Alice, it doesn't. Alice owns Bob^WEve's computer, conveniently named Bob; Eve^WAlice wants to talk to Bob, but that silly Bob^Eve who actually owns Eve^WBob (in Bob^WEve's world, not Eve^WAlice's) wants to eavesdrop on the conversation between Alice and her computer Bob which Eve just leased.
Clear now ?
You recently learned wrong, however. It doesn't matter whether DTAG or Deutsche Post are private enterprises or not; article 10 still applies. Without a law allowing it or a corresponding court order, government officials still cannot go to the Post, open your letters and read them. Article 10 still prevents them from doing that. The German state remains a substantial shareholder in both DTAG and Post, as well. Also, even the private enterprise Deutche Post (and its competitors) are bound by the Postgesetz (as is the DTAG and its competitors by similar laws) which codifies pretty much the same -- i.e. even the private enterprise Post cannot legally read your mail or collect information on your mailing behavior. (See 39 PostG, for instance).
... Analysis of user behavior is not a valid reason to log personally identifiable information as per the law.
Furthermorse, it's not article 10 that is in question here -- it's the BDSG. We don't need case law for what you describe to still be illegal; the BDSG is pretty clear on what it considers personally identifiable data. While no case law exists, judges are still free to look at other decisions and interpretations.
"I'm pretty confident that 'regular' logging will continue to be alright; the analysis of user behavior is the critical fact here, at least that's how I read it." -- how you read what ? Have you READ the BDSG ? Have a look at http://www.gesetze-im-internet.de/bdsg_1990/index.html
There's an old saying "Wo kein Kläger da kein Richter", though. People don't usually sue Joe Sixpack Webmaster about this stuff.
Some laws proposed recently are silly, I agree. The BDSG is not one of them. The now infamous Hackergesetz, however, is.
Your logic is fallacious.
A single IP address is not necessarily associated with a single person. Correct. A -> B. This does not imply B->A in any way, shape or form.
The site actually doesn't make that argument, however. It makes the argument that an IP address is not permanently associated with a single person and easily changed for most (most ISPs here assign you a different IP on each login, out of a pool of millions; and most ISPs here do not allow connections to stay connected for longer than 24 hours).
Furthermore, the site states the exact opposite of your assertion a few paragraphs later. IPs are, in fact, personally identifiable to at least the government, police, and intelligence agencies (as well as foreign hostile intelligence agencies and witty hackers of the legal and technical kind) since ISPs store that data (even though they are not required to (yet) and actually currently forbidden to, lawfully.
Last, but not least, your jump from "it's not exactly 1 person == 1 ip" to "it's not personal data at all" is plainly wrong. Take phone numbers as an analogy. You can clearly change phone numbers. Are they suddenly less not associated personally with you, AT ALL ? Take credit card numbers. You can have many of them, or share one with several people, or even change them once they become compromised. Does that make them any less personally identifiable ?
Funny, if only it were true. The law forbids the general collection of all traffic data without due cause. It does not forbid personally identifiable logging on a case-by-case basis to resolve problems and issues.
To sum up ... Will somebody PLEASE think of the children ? :P
... Of course you can evade problems in this area by effectively anonymizing the data in a way that cannot be reverted even given auxiliary data -- once that is done, the data is no longer personally identifiable and does no longer fall under the strict privacy laws.
An argument can be made that IP addresses you use during surfing are data that can be tied to you personally, and as such fall under the strict privacy laws we have here. Don't like it ? Change the law, or work within it. (of course next to nobody actually cares about that law -- the logging you refer to at the ISP level is just as illegal, but has nonetheless been happening for years. DTAG, the biggest German ISP, logs customer IP addresses in violation of the data privacy laws -- they have been sued for it, and at least one customer no longer has his IP logged. Well, only that one customer, that is. Everybody else is still logged.)
You make a valid point about balance. There might still have to be a distinction between government-provided services (such as the site in question) and privately owned enterprises (equal under the law to you and me, while the government is in a clear position of power). In fact, there is; As a citizen you have considerably more power against the government than you do against private enterprise (since the Grundgesetz (our constitution) applies to the disparate legal relationship between the citizenry and the government, not between citizens as such (which corporations are a part of); the most powerful legal tools are reserved for these battles; even if the (strict) privacy laws do not apply, you can still sue based on article 2, for instance (which guarantees the right to personal freedom, an umbrella which includes the right to decide what happens with personally identifiable data about yourself).
You also seem to throw together the laws and practices of the US and Germany into one pot. I'm not certain every phone call ever made is logged here. I'd be surprised if it were, especially when flatrate-type fees are involved.
Personally I prefer tough privacy laws over mandatory data retention and the assumption of general suspiciousness for years. Wiretaps can still be ordered by the court. IMHO, the ISPs and Telcos should not be doing the policework FOR the police. It's not their job.
Web admins can still effectively manage their servers. They may not be able to log IPs for extended periods of time, but they can STILL manage their servers. They can still block IPs. They can still determine the origin of attacks as they are happening. Evidence can still be collected. However, log analysis and data retention for extended periods of time may become shady
Indeed it should. It is not, however, a security issue, and as such it falls under the system administrator's job to ... you know ... administrate his or her system.
Nobody is screaming for an update to Slackware 1.0, either. Same thing.
There are many reasons that airports have special significance:
...) What if there's just a regular accident, no terrorism involved ? The same things can happen. It doesn't really make the airport special and holy. The terrorists may have to find a different target, but a target they will find. Those batshit crazy buggers aren't always without intelligence.
One by one, eh ?
Terrorists have a track record of targeting them
Also, Skyscrapers, Pizza places, night clubs, crowded areas, buses, cars, subways, college campuses, palaces, houses of the wealthy, and no end of other things. You see, "terrorists" are like snowflakes. None are alike. Some batshit crazy terrorist will find something you'd find to be without any special significance significant enough to blow up. Be afraid. Be very afraid.
Come on, that doesn't even come close to justification or counterpoint.
A small bomb in a park can kill a few people. A small bomb on an airplane can kill a few hundred.
A small bomb in a space station can kill a few less people. A bomb in a park while there's a music festival there can kill more people than there are in many planes. A bomb in my backyard can kill a squirrel. A bomb in a bus can kill slightly less people than in a plane, slightly more than in an empty park.
What, exactly, is your point ? What makes the airport special ? Bombs can kill people. We established that. How is killing people at point A somehow more significant at point B ?
(and note that you can probably detonate a minivan full of explosives in a park, while you'll probably have less thereof on a plane. If you really want to optimize your terrorist attack for the variable X, where X is number of people killed, and X is to be equal to some set number, you can probably figure out a plan to do just that, use just the right kind and amount of explosives, etc.
If you set off a bomb at a gas station or a grocery store, people will still buy gas and groceries. But most business travel and all tourism is discretionary. So one bomb can result in the loss of millions or even billions of dollars of economic actitivity.
You are elevating fearmongering to an art.
Let's go into a hypthetical. Let's say we set off a bomb at a walmart located close to the White house, and then another 5 minutes later at a gas station on the other side of the white house. Mix in some napalm so that it really burns for a long time and is nicely visible. What kind of impact, do you think, would such a thing have if there happened to be a couple of news networks at the scene within seconds, without much other information ? PANIC ! PANIC ! Washington is under attack ! Let's dig in and only come out when the president says it's OK !
Well placed bombs in locations other than airports can easily disrupt the economy. Hell, meaningful terrorism of any kind can do that. Planes are not special in that respect. What if you blow up a train ? (make it contain some radioactive waste, just for kicks
None of your arguments actually show why an airport is more significant than a busy bus or train station w.r.t. the amount of force, searches, and general suspicion. Heck, my laptop bag contains some scaaaaaary electronics (i.e. home-built and not necessarily encased). I'm looking forward to explaining how it all works to the hopefully friendly guy pointing a gun at my face, one day.
... you know, continue using the GPLv2 for your software. Nobody is forcing anybody to use the GPLv3.
Much ado about nothing. Come on, be a little more creative with your FUD.
Let's knock out the obvious ones first, shall we ?
/can/ use this as a storage-only service, but of course you can also use it for distribution and such -- no transfer limits. Rapidshare.com has similar offerings (with "unlimited" storage but a 5gbyte/day cap) at similar prices. Both of these rely heavily on customers infringing copyrights, so it's anyone's guess how long they'll stay around. Both also use somewhat nonstandard file deposit and file delivery methods. There are countless others in that market (oxedion, mediafire, upfile, rs.de, filefront, etc., all with varying foci).
...
http://www.megaupload.com/ has one offering, 250gbyte. Prepaid for one year it's 50 Euros (or whatever their site says for the US locale). That's 70 bucks. You
The regular webhosting market has things like this to offer as well. http://www.dreamhost.com/ : The cheapest plan, at one year prepayment, would be around 120 bucks and offer 145gb of space. I say would since you can use their promo codes (check the forums) to almost triple the space or drop the price to a lot less. So that's 400 gb of storage, a couple terabytes of transfer a month, and some processing power to boot (WebDAV/FTP/SFTP/SCP/rsync/etc. are all possible). I imagine competitors to DH will have similar offerings space-wise. We're looking at around a fourth the price for almost double the storage space. Don't you dare yell "overselling" -- Google does, too.
If you can be bothered with some cumbersome setup (to laypeople, anyway), Amazon S3 will get you storage space for $0.15/gb/month, plus traffic ($0.18/gb). If you actually use 250gb, the price will be comparable to Google for storage alone (i.e. no transfers other than the initial incoming transfer); the difference is that you get charged by the byte, not in large pre-paid packages. If you use 1gb and transfer it twice, you pay $0.51 that month. Also consider that if you use less than the 250gb Google offering, you're probably get away cheaper (since the smaller Google plans are comparatively more expensive while Amazon's offering exhibits a linear price curve over the amount of storage used).
The value Google's space has is probably the integration with its applications -- Picasa, for instance, lacks decent online functionality using standard protocols -- and Google will probably deliver GREAT online functionality with their own service.
If all you really need is a foolproof backup, open up an FTP and let the world mirror it. I wonder who would do such a thing
No, really. If you don't actually mean "it's free for whatever you want to use it for", don't release it under a license saying that. It's quite simple, really. Bitching and moaning after people take you up on your offer is just childish. Sure it would be nice of Sun to help the project financially. Same goes for Cisco. But you damn well better don't EXPECT them to. If you give a gift to the world, it is exactly that. A gift. No strings attached. If you want strings, freaking spell it out in the license. In that case it's not a gift anymore, but a product for sale with you expecting compensation. Maybe that's why they charge for their install disks.
You clearly know nothing about OpenBSD. Just because I don't agree with everything they do 100% or wouldn't follow Theo everywhere does not imply my lack of knowledge about or usage of OpenBSD. But hey, it makes for a cheap ad hominem attack, doesn't it
TrueCrypt, while a fine piece of software, could never offer what this stick APPEARS to offer -- namely data eradication in the case of an attacker attempting a wrong password too many times in a row. This would protect against brute-forcing the password, possibly on a cluster of computers.
TrueCrypt has been broken before, just not in the way you believe. The methods they use are sound, but no amount of secure encryption can protect you from a brute-force attack on weak passphrases. The idea with the secustick is that you don't even get access to the ciphertext without credentials being present (and a truly decent stick would encrypt the data with the given password anyway, making the debugger a useless tool against it; people could possibly still disassemble the stick and try to dump the memory which would give them the ciphertext to attempt brute-force attacks on, but that's another barrier to break -- especially if the stick is built in a way that would hamper efforts to disassemble it without destroying the data.
Ah well, there are enough knuckleheads who will buy anything that says "Secure ! We promise !" without actually understanding whether what they're buying is snake-oil. Capitalism rocks, that way.
You can change the meaning with adjectives. That has not happened here. How is it abuse if I have the tube running 24/7 as background noise ? Don't assume that just because YOU can't imagine a way to keep your TV on 24/7 there isn't anybody who might legitimately do so. Hey, maybe they are too lazy to press the off button, or they'd rather mute it instead. That's not abuse. And let's not forget that the cable company can easily throttle back your general Internet bandwidth in the case that you're using excessive IPTV bandwidth. Unless, say, the IPTV is not delivered by the cable company but by a third party. Whooops. Triple play and all that crap sound nice on paper for the cable company, customers expect free choice of competitors, though. Now as far as the people in the article, they are CLEARLY using their cellphones as a general Internet connection for their computers. This is FORBIDDEN by the cellphone TOS unless you sign up for a different plan. I just flat out don't believe that someone used 5GB of bandwidth in a month by checking email and surfing web pages using ONLY their cellphone. You know that emails can easily be 100m or more in size ? Webpages can contain media ? "surfing webpages" is not a clearly defined term ?
You also forget that intranet access is perfectly acceptable. My intranet has tons of data to be perused. Note that " downloading legally acquired songs" (term from the ToS) is indeed allowed. Whoopdidoo, FLAC is big these days.
"Anyone using more than 5 GB per line in a given month is presumed to be using the service in a manner prohibited above,"
Is quite interesting as well. It's a blatant lie, all around. They could just as well have written
"Anyone using more than 5 GB per line in a given month is costing us more money than we alotted for this 'unlimited' plan, so we cut them off."
At least that would have been honest.
> IPv6 is fully integrated.
r tnonexistantduetomissingprotosupport-0728.
:)
Oh really.
Maybe so, they just kinda, sorta forgot proper IPv6 tunneling options. You know, the stuff you need to get it going on the current net infrastructure. Also the stuff that Windows XP does without a hitch. See
http://www.sixxs.net/news/2006/#windowsvistasuppo
If you want to use tunnel brokers, you'll have to install third-party VPN drivers (OpenVPN). This is a feature that SHOULD have been there from the start if Microsoft cared about full, usable IPv6 integration.
> There may still be people who like Windows 98, but there aren't people who use Windows XP, and say "Gee, I wish I was using 98 instead."
I know a few people like that. Windows 98 still runs better in low-memory situations, does everything "they" need, and quickly at that. 2k/XP are better operating systems overall IMO, but gee, they are not the hammer to every nail
> So shall it be with Vista when it matures.
So operating systems are kind of like wine ? What if Vista is a bad year ?
You write : "If you are going to use math to make an argument, you damn well better get it right. .1831 GBps, or 10.986 GB/min, or 659.16 GB/hr, or 15819.84 GB/day, or 474,595.2 GB/mo."
.0003 USD, but 0.003 USD. (and even .0003 USD would be wrong if you assumed terabytes instead.) All the rest of your math is pretty much wrong due to these rather grave mistakes.
Hmm, let's see. A typical T1 line delivers data at the rate of 1536 Kbps (don't bother about the extra 8Kbps, OK?). So, that's 1536000 / 8 / 1024 ^ -2, or a whopping
How you get to 1536000 / 8 / 1024 ^ -2 to 0.1831 GBps, God only knows. Bits or Bytes, 10^-2 or 1024^-2, etc.
KISS. 1536kbit/s = 192kbyte/s. A month has 30*24*60*60 = 2592000 seconds. 192*2592000 = 497664000. So you use 497664000 Kilobytes per month. That's 474 Gigabytes. To arrive at your claimed number, we'd have to use 1000 T1s or 1.5 gigabit/s. That is a tad more expensive, even if we are not talking about the '90s. Also note that in networking, usually we deal with powers of 10 instead of 2.
So it's not
You also fail to realize that you don't buy gigabytes, but bandwidth. More to the point, the provider and its backbone most certainly don't consider gigabytes per month, but bits/second of bandwith. A 1.5mbit/s line does not mean that you actually transfer 474 gigabytes a month. It just means you have the capability to should you be able to continually fill it. Thinking of this in terms of gigabytes is just easier on the (usually somewhat dense) marketing department, and is a lot easier to convey to stupid prospective customers. Let's face it, how many of em would understand 95th percentile billing for burstable bandwidth, or even why that's not such a bad idea ?
Last, but not least, maybe some more data. You can get 1mbit/s of bandwidth with 95th percentile billing for $40. Some are cheaper, some are more expensive, but that's a good baseline of what you can get in a datacenter. If you cap your bandwidth at 1mbit/s, that 95th percentile is also 1mbit/s, so that's valid. Rethink your calculations now, and take into account that a DC does not have to roll out the last mile, unlike ISPs.
ISPs are still usually corporations and want to make money, some are greedy, and many could offer their services cheaper. Obviously if it really was that much cheaper as you claim, go compete with them. I mean, come on, if you can do it 10 times cheaper, just halve their prices and still make 5 times as much as you spend ! So easy, eh ? Thought so.
SELECT * FROM bad_analogies WHERE content LIKE '%grep%'
You were saying ?
Asiding the problem that BitTorrent interclient communication is generally not UDP, this would also increase the tracker load manifold. Currently, each BitTorrent client contacts the tracker every 5-60 minutes or so. Using this technique could quite possibly increase this to a few packets per second per client on busy swarms (assuming, for the sake of simplicity, 1 pps per client, that would result in a 300fold to 3600fold increase in pps on the tracker. A tracker of ~50000 peers currently serves 300-350 requests per second. Increasing that by a few orders of magnitude is not exactly high on a tracker operators' agenda (though it might be worth it for corporate content distribution systems).
As a sidenote, SIP implementations (internet telephony) use the STUN protocol to much the same effect.