Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
United States

Bush Wants an Unhackable Private Network 365

Slur points out an article at the New York Times which says that the "Bush administration is considering the creation of a secure new government communications network separate from the Internet that would be less vulnerable to attack and efforts to disrupt critical federal activities," writing "It seems to me money would be better spent getting the next-generation Internet going, for the government to fund more of the existing research and standards boards to create protocols that are invulnerable to the kinds of attacks the government seems to fear, namely massive DOS attacks. Or is there something else a 'net terrorist' could do to 'disrupt the vital flow of information'?" Isn't hard-to-disrupt communication the reason that DARPA got involved in this "Internet" business anyhow? Update: 11/19 22:48 GMT by T : This was mentioned before a little while ago when USA Today wrote about the same concept, but apparently a Digital Pearl Harbor is still being flogged.
This discussion has been archived. No new comments can be posted.

Bush Wants an Unhackable Private Network

Comments Filter:
  • by st. augustine ( 14437 ) on Monday November 19, 2001 @06:31PM (#2586773)
    Bruce Schneier has an informative story about this in the November 15 CRYPTO-GRAM [counterpane.com], including some of the pros and cons. Basically, he says it would be better than what they have now, but still not all that great (he points out that the government already has several separate, secure internets, for various purposes, and they were still infected by Melissa and LoveLetter). And that this is one of the few cases where security and convenience might really be inversely proportional.
    • by Philbert Desenex ( 219355 ) on Monday November 19, 2001 @07:03PM (#2586961) Homepage

      the government already has several separate, secure internets, for various purposes, and they were still infected by Melissa and LoveLetter

      Now that's something we didn't see on C|Net.

      I worked in the aerospace industry from '86 to '92. Every big defence contractor had one or more classified IP networks. Unfortunately, the security measures imposed were sort of stupid: the ethernet cables of the classified net had to be at least so many feet from a phone line (they were worried that induced voltages from ethernet would allow someone on the phone to "tap" the classified net), keyboards attached to computers attached to the classified net couldn't be traded out to unclassified areas, and had to be elaborately destroyed when they broke. At the same time, you could walk through checkpoints with pockets full of floppies.

      It was as if a Korean War Drill Instructor dreamed up ways to actually impede using the classified network, but at the same time allow (possibly) classified information in and out of the building.

      • RE: Unfortunately, the security measures imposed were sort of stupid: the ethernet cables of the classified net had to be at least so many feet from a phone line (they were worried that induced voltages from ethernet would allow someone on the phone to "tap" the classified net)

        This is actually true. You could and do get enough crosstalk that a good sniffer in van could pull packets off your ethernet.

        RE: keyboards attached to computers attached to the classified net couldn't be traded out to unclassified areas

        Maybe they're worried about trojan hardware? A keyboard gets borrowed out, a small modification is made so that it logs every key pressed and then a week or two later gets "loaned" out again to extract the data.

        remember these are people who get payed to be paranoid.
        • This is actually true. You could and do get enough crosstalk that a good sniffer in van could pull packets off your ethernet.

          You'd have to explain why the building where this classified network resided had offices with glass windows, and terminals ('92 remember?) facing the windows. The "security" people apparently didn't consider someone with a telescope a threat.

          Maybe they're worried about trojan hardware? A keyboard gets borrowed out, a small modification is made so that it logs every key pressed and then a week or two later gets "loaned" out again to extract the data.

          Let's see... keyboard gets used a maximum of 12 hours a day, and an engineer types 50, 5-letter words a minute. That's 12 x 60 x 50 x 5 = 180,000 bytes of info a day to store in the keyboard. Nope. Even in '92, we had 1.44 Megabyte floppies. It would have been much more efficient to move info via floppy. Security folks being dumb again.

          remember these are people who get payed to be paranoid.

          You make a correct statement, but "paranoid" doesn't mean "intelligent". It means "a variety of insanity". I'd rather have security people paid to be intelligent, than paid to be insane.

  • question (Score:2, Insightful)

    by Anonymous Coward
    :Isn't hard-to-disrupt communication the reason that DARPA got involved in this "Internet" business anyhow?

    Yup
    • yeah, it was hard to disrupt w/a nuclear explosion taking out half the country yet it isn't hard to take out a good majority of the network now by sending around a DoS attack that spreads.. A nuclear blast was theoretically a localized event (although a limited engagement is something that is debated). A DoS attack (as has been shown) spreads fast and furious due to stupid people not protecting themselves. Lead walls won't protect Lisa this time...
    • Re:question (Score:2, Insightful)

      Yeah, but we're talking about completely different kinds of disruptions here. The APRAnet was designed to resist machine failure at critical hubs, caused, for instance by them being blown the hell up.

      It was NOT designed to be secure to attack from the inside--and with the global Internet, everybody is inside now.
    • answer Re:question (Score:2, Informative)

      by gilroy ( 155262 )
      Blockquoth the posters:

      Isn't hard-to-disrupt communication the reason that DARPA got involved in this "Internet" business anyhow?

      Yup

      Um, nope.

      While some work had been done on using packet-switching to improve communication reliability after a nuclear attack, that work was purely theoretical and not directly tied to the origin of the ARPAnet. The ARPAnet was explicitly created to allow computer researchers to share files and resources, reducing unnecessary duplication of effort and resources. The nuclear war myth might be better copy, but it's just a myth.


      Check out Where Wizards Stay Up Late for the real story.

      • by man_ls ( 248470 ) on Monday November 19, 2001 @07:03PM (#2586959)
        According to The American Institute of Physics [http] in their Physical Review Letters journal article "Resilience of the Internet to random breakdowns" (19 Oct 2000) [a copy of this article is available in .pdf from my personal web page [tripod.com] on the left side bar for your reading pleasure.] stated that the Internet could lose 99% of its nodes, and still maintain routability. The content lost in those 99% of nodes is another matter, but the Internet would not segment until over 99% of the routing nodes were removed. That's pretty impressive.
    • Re:question (Score:3, Informative)

      by Alien54 ( 180860 )
      Isn't hard-to-disrupt communication the reason that DARPA got involved in this "Internet" business anyhow?

      But somehow that all went to hell when it got commercialized. How many people here remember the splash made by that first infamous piece of broadcast spam from that lawyer in Arizona?(or was it California?) Or the September that never ended with the advent of Internet access via AOL.

      As soon as all these commercial interests got into it, wham. And this is the information superhighway invented by algore. The bloody mess of spam and commercial jerks. Not Darpa

      • Think about it: when the Internet was restricted to non-commercial nodes, it was pretty secure. The first major security disaster was the Worm of 1988, which came from a university site.

        If you maintained a separate TCP/IP network that only had physical connections on military bases and the like, I'd think it would be pretty secure. It's this business of giving everybody an Internet connection that gets all the script kiddies online.

    • "The internet was designed so that, in times of nuclear war, the United States Military would have free and east access to pornography."

      dave
  • Isn't this a repeat? (Score:2, Informative)

    by Krimsen ( 26685 )
    Wasn't this covered [slashdot.org] back in Sept?
  • Already exist (Score:5, Informative)

    by firewort ( 180062 ) on Monday November 19, 2001 @06:32PM (#2586787)
    Bush may not know it, but these already exist in the form of SIPRNET, and INTELNET.

    SIPRNET

    SECRET INTERNET PROTOCOL ROUTER NETWORK

    SIPRNET will replace the DSNET-1 during the migration to DISN. It operates at the SECRET Collateral level and can interface with the TROJAN network. It provides higher and selectable data rates at a much lower O&M recurring cost. Inter-site data rates are 512 Kbps and in some cases T-1. Users can connect to the network at selectable data rates that meet the need.

    INTELNET

    NAVAL INTELLIGENCE COMMUNICATIONS SYSTEM

    The NICS is designed to consolidate Naval Intelligence communications systems. The system has three parts. INTELCAST plan calls for each FOCIC or Facility to consolidate up to 12 different message traffic circuits, including OPINTEL, MUSIC, FIST, and DODIIS through INTELDATA extended in an SCI LAN Extension and Stand Alone capability configuration. The SCI LAN encompasses a full suite of SOCRATES equipment, including workstations, secondary imagery dissemination systems, and a mapping and graphics capability. The Stand Alone capability provides a workstation with tailored data bases specific to unit operational orientation. Stand Alone capabilities are being provided to Guard and Reserve units as well as to certain active, lower-echelon units.

    NIPRNET

    UNIFORM INTERNET PROTOCOL ROUTER NETWORK

    The NIPRNET is the consolidation of several service/agencies networks (AFNET, NAVNET, MILNET) with common protocols and standards. It is a product of the DISN near Term Program, which sought a reduction in cost of operation through interoperability and standardization. Connectivity over high-speed trunking is supported by the NIPRNET. It operates at the unclassified level, while the SIPRNET supports classified networks in a similar manner.
    • and can interface with the TROJAN network.

      It's definitely much safer to input and output if you're interfacing with TROJAN :)
    • Works well right?
      Until part of it goes down again like it did last month (sept) and you have to use secure faxing right?
    • From MARKING CLASSIFIED EMAIL MESSAGES ON SIPRNET

      (Original all caps, lameness filter encountered)
      Until an automated solution has been evaluated and approved for use in the USMC, classification markings will be done MANUALLY.

      "Um Sarge, when can I clean all these ink stamps off my monitor"
    • by tcc ( 140386 )
      >Bush may not know it, but these already exist in the form of SIPRNET, and INTELNET.
      >SIPRNET
      >SECRET INTERNET PROTOCOL ROUTER NETWORK

      Ok It's a secret, Shhhhh! only you and 2,000,000 more readers now knows about it :)
    • by budgenator ( 254554 ) on Tuesday November 20, 2001 @08:23AM (#2588533) Journal
      Remeber JINTACCS? I doubt it, it was a messageing system, actualy kinda like XML. It allow an Army soldier to do things like call it Naval gunfire. On the lowest level it was a fill in the blank paper, then read over voice radios, at the higher levels a computerized intercomunications protocol.

      Actualy it was a good system, not perfect but good, but it was murdered. They did this by teaching it. They didn't start with the easiest and work to the hardest, they tought the hardest first so the average pvt Joe Snuffy got hopelessly lost. They actualy tought me how to report the laying of a naval mine field, I was in an light infantry organisation at the time, that report was for Naval ships Captains. This happened because the middle management types realy didn't want to lose their turf. I think the same thing is going to happen here.

      To us its easy, blow some fiber, install some routers between facilities, gateway to some secure sattalites and maybe change the networking code enough to make the civilian stuff incompatable. Add in an armor plated authetication, distr the software to authorized users and your done right? Well the Army won't like working with the Marines, DOD won't like working with DOJ, and Intell won't even like working with themselves.

      The only good thing I see from this is sonner or later some of the reasearch is going to trickle down to us and be usefull.
  • In the beginning (Score:5, Insightful)

    by Dirk Pitt ( 90561 ) on Monday November 19, 2001 @06:34PM (#2586798) Homepage
    It seems to me money would be better spent getting the next-generation Internet going


    It seems to me this would evolve just the way the Internet did before; it would at first be used just by government agencies, next given to the large defense contractors, eventually adopted by the research universities, and then swallowed whole by Joe Public. This, IMHO, is the best way to get the next-gen Internet.

    • Blockquoth the poster:

      It seems to me this would evolve just the way the Internet did before; it would at first be used just by government agencies, next given to the large defense contractors, eventually adopted by the research universities, and then swallowed whole by Joe Public. This, IMHO, is the best way to get the next-gen Internet.

      This might well be the evolution of this new network, but it is not how the current Internet evolved. The Internet, as ARPAnet, was explicitly for the research universities from the get-go. The first nodes on were universities; the first "commercial" node was BBN, the consulting firm charged with building the net.


      The government, in fact, was in general quite reluctant to get into something that was perceived, at best, as a convenience for computer researchers.

  • Grow up, Georgie (Score:2, Flamebait)

    by babbage ( 61057 )
    "Bush Wants an Unhackable Private Network"

    And I want Bambi's father to come back, but it ain't gonna happen. Sorry to disappoint you with this Real World stuff, Dubyuh, but there's no such thing....

    • by Xerithane ( 13482 )
      Feel free to hack into my home network. It's IP range is 192.168.0.1 - 192.168.0.13.

      Running drywire or some other method of lines as long as they are physically seperated from the rest of the internet (think of the way the bank systems do this via verifone boxes) does make it unhackable and private

      Of course, it relies upon physical security and not so much bit-based security. Before flaming our president understand it is a real concept. And I'm sure he has quite a few people that know a lot more than you do on the matter; never try to know everything just know people who do.
      Note, he didn't say an "internet based private unhackable network" but a private network. My guess in the private IP range. Considering all the secure channels (via satellite, or some other method of communication) I'm sure that this can easily be achieved. Granted all that, I do think it's a stupid idea... but realistic none-the-less.
      • by dougmc ( 70836 ) <dougmc+slashdot@frenzied.us> on Monday November 19, 2001 @06:55PM (#2586928) Homepage
        Feel free to hack into my home network. It's IP range is 192.168.0.1 - 192.168.0.13.
        Already done. My login and password are so ubitquious that they work on these systems as well!

        Alas, they don't seem to have any mp3s or warez that I don't already have. Bummer.

      • Re:Grow up, Georgie (Score:4, Informative)

        by Cally ( 10873 ) on Monday November 19, 2001 @07:28PM (#2587047) Homepage

        Feel free to hack into my home network. It's IP range is 192.168.0.1 - 192.168.0.13.


        How wonderful, someone who still thinks NAT equals security!

        I'm not going to spell it out to you, but I suggest you:

        1. tighten up your firewall rules immediately. (You ARE running
        a firewall, aren't you?)and

        2. Start checking your IDS logs closely for the next few days.
        (You ARE running an IDS, aren't you?)


        OK, if you want further hints for your googling: firstly, look for `arp poisoning Dug Song MitM'. Then search the Bugtraq, and perhaps the sec-focus Pen-testing list archives, for info about how to own the OS/platform you're NATing with (ie if you're NATing thru Linux, I mean the Linux box.) Remember to check for known vulnerabilities in the services that show up when you nmap your external interface. Yeah, of course you're completely up to date with all current patches, but I bet that there was a window of vulnerability before you applied each one...

        In general, boasting on Slashdot about how secure one's network is, is a BAD idea.

      • I understand that it is a real bad concept. (Kinda like missile defence, but that's a whole other flame war... :). Go read the Bruce Schneier article that was mentioned elsewhere in this discussion, then reconsider your position. The value of a network rises as the number of nodes rises, and as a corrollary falls as the number of nodes falls. Thus for this private government [contradiction in terms?] network to have value, it will have to be big enough to be of value. But as the size of the network increases, the difficulty of defending it also increases. And the difficulty of having a sizable network that really is completely physically separate from the public internet will be considerable.

        Think about it: every employee could end up needing two separate computers on their desk, one for the local network and one for the government one. That employee would have to be vigilant about not ever transferring files from one to the other, either by wire, wireless, or disc. If the employee needs to transfer an email, it'll have to be a hard copy or a retype. If any personnel have laptops, they can't be brought out onto the internet, and laptops from home can't be plugged into the network. For that matter, pretty much any kind of wireless networking is out since none of it can be trusted not to accidentally send or receive anything that wasn't supposed to be sent or received.

        The chief problem here is that it places a ridiculous emphasis on perimiter defence without paying any attention to internal defences. Kinda like missile defence. Kinda like a bad firewall product. Kinda like the Maginot Line. These kinds of systems are difficult to set up in the first place, difficult to maintain across any span of time, and once a chink in the armor is found you tend to have a complete collapse in defences, because you've placed all your resources into this one point of failure.

        Again, read the Schneier article, and the points about viruses running rampant through military networks because some idiot plugged his laptop into both the public & private networks. If this proposed network is to be useful, again, it will have to be big -- because the utility of a network generally rises as the square of its node count -- but chances are the difficulty of defending it will rise at about the same rate. That's untenable in the long term.

        You're right that I'm no expert, and maybe the people advising the moron in the white house are smarter than I am. Certainly they were pretty clever to get that Orwellian Patriot Act passed without anyone noticing in time. But my hunch is that if we want to have some sort of secure networking capabilities, the way to do it is not "vertically" by cutting off parts of the 'net & placing them behind a Maginot line, but "horizontally", with secure protocols, encryption, and the like. I'm not well versed enough to express this more coherently, but it seems to me that protocols like ssh are reasonably secure while being able to leverage the high utility of a large network, whereas this kind of isolated subnet can't guarantee any greater level of security and yet it loses out on that large network usefulness.

        • Think about it: every employee could end up needing two separate computers on their desk, one for the local network and one for the government one. That employee would have to be vigilant about not ever transferring files from one to the other, either by wire, wireless, or disc. If the employee needs to transfer an email, it'll have to be a hard copy or a retype. If any personnel have laptops, they can't be brought out onto the internet, and laptops from home can't be plugged into the network. For that matter, pretty much any kind of wireless networking is out since none of it can be trusted not to accidentally send or receive anything that wasn't supposed to be sent or received.
          Not really. You simply use an encrypted VPN between the Internet/Dubyanet interface and the workstation.

          Security could be implemented, say, with a one-time pad that is keyed to the workstation actual address (so if the key is stolen, it can't be used elsewhere to spy on the conversations).

    • If you can hack into a separate physical network than the general internet good luck. And there is hardware encryption encrypted with more hardware encryption much stronger than the measly 128 bit that us civilians use. If I remember correctly someone told me it was something like 1024 bit at the lowest level.
      • I just can't parse the beginning of your second sentence. There is ...what, exactly? Hardware encryption with more hardware encryption? I don't know what that's supposed to mean...

        Anyway, if you see a very tall fence that goes part of the way around the building, do you try to go over the fence, or do you try the gate? Hacking into this network from home may well be an exercise in futility, but that isn't to say that it'll be safe from malicious or incompetent insiders.

        And key length really doesn't mean very much. A long key with a bad encoding algorithm is no better than a short key with a good algorithm, or put another way, if that 1024 key chain runs an algorithm that can only generate 32 bits of entropy, then you might as well just use a 32 bit key. Furthermore, keys of the same length aren't necessarily of equal quality. A clever algorithm might be able to get more use out of say 40 bits than a less clever algorithm does in 64, but then that's just the earlier idea expressed in reverse.

        In any event, the main point is that key length looks good in marketing literature, but the best way to know for sure is to have a cryptographically established algorithm, and the more open that algorithm is the better you can trust that it's actually secure. Don't be impressed just because someone told you an algoritm can spit out lots of bits, since anyone can do that:

        for (1..10000) { print $_; }

        Hey look at that I just came up with a ten thousand key algorithm, I'm smarter than the NSA! Yeah right... :)

        • by alen ( 225700 )
          The military has been sendding encryption keys over the radio waves for years. Naturally it has found a way to encrypt them. As far as my post here is what someone told me before an exercise I helped set up. The intel people's data is classified top secret and is encoded with the appropriate encryption. General classified data is secret and isn't encrypted as well as top secret data. At another point these two streams are combined with plain text data and then encrypted again. The opposite happens at the other end. Here is some info on the web: KIV-7 [rainbow.com]

          KG-84 [fas.org]

          Secure telephones [tscm.com]

          The NSA has some really smart people to rip this stuff apart and certify it to be secure before it goes into production. These products are usually designed to a higher standard than software programmed by people in their spare time or microsoft.

          • Long key length doesn't mean hard to break. Overly complex encryption schemes doesn't mean hard to break. I'm sure these people are very smart, and I wouldn't pretend to have a clue how to break them myself, but the fact is that it's silly to say that any encryption strategy is strong just because it's impressively arcane. The fact is that for regular personal & commercial use, ciphers of as little as 128 bits are perfectly safe and will remain so for a good while -- distributed cracking efforts don't really invalidate them as much as they prove how difficult they are to break, and they have proven that they are in fact comfortably difficult to break. I'm sure the NSA wants a higher level of comfort, and I'm sure they have a lot of smart people that spend all their time trying to do even better, but I'm also sure that anything that is cryptographically secret or proprietary is also cryptographically unproven. That might be okay -- the NSA might not be too worried about formal academic proofs for all I know -- but in the absence of better knowledge and analysis, it's really impossible to comment on the quality of what they're using.
  • by Anonymous Coward on Monday November 19, 2001 @06:38PM (#2586815)
    Wouldn't creating a wholly separate network for restricted traffic be a bit counterproductive?

    I mean and spy/hacker who found a physical location to hack into it (i.e. tapping into a line on a phone pole or at a phone company switch) would find *everything* on that network to be of interest. In essence they would have hit the jackpot for illicit information. We're kind enough to organise it away for them.

    True it would probably prevent 15 year old script kiddies from casually hacking in at home, but it would make any break into that 'other' network all the more catostrophic prospect.
    • These aren't like networks you have probably ever seen though. The current government "secure networks" aren't VPN's or anything. They run on their own lines between very secure (heavily guarded, extremely redundant security) data centers (ie. DMS has 2 in europe, 2 in the pacific, and like 10 in the USA). The traffic between data centers is encrypted with proprietary DoD software. From data centers to the end user, data is encrypted (once again, with proprietary software) and is read using an off the shelf e-mail client. So, for your lucky spy/hacker to really hack the network, he/she would have to hack either the Encryption for which he or she will never be able to find the algorithm, or just hack the computer of one user. Even then though, the hacker would only have one side of the communications and most of it would probably be of little interest as the DoD uses a 7-12x random overwriting scheme to destroy sensitive computer data. Intercepting transmissions between the user and the data center might be interesting, but still this is a Departement of Defense Computer. I think they keep pretty thorough logs and any exploit would be quickly terminated.
    • yeah, well, that's the idea...

      put all your eggs in one basket... and then WATCH THAT BASKET!!!

      eudas
  • by Cally ( 10873 ) on Monday November 19, 2001 @06:39PM (#2586825) Homepage

    the kinds of attacks the government seems to fear, namely massive DOS attacks. Or is there something else a 'net terrorist' could do to 'disrupt the vital flow of information'?


    The problem is that much of the 'vital information' in today's society flows over the public internet - by definition. Sure, take military command and control comms out of band - that makes perfect sense anyway, which is probably why there are several separate, highly secure military and governmental IP internetworks that are supposed to be completely separate from the public Net. (Although, as Bruce Schnier points out in the latest Cryptogram [counterpane.com], ILoveYou made it onto the 'secure' network within 48 hours...
  • by Ieshan ( 409693 ) <ieshan@g[ ]l.com ['mai' in gap]> on Monday November 19, 2001 @06:40PM (#2586827) Homepage Journal
    What he's asking for is like asking for poison-free food. Sure, the ovens can be locked and the food can be tested over and over, but the cook is still there.

    The only concievable way to do this is to either:

    a) Eliminate Government Data Access to All But the Highest Officials (which still poses the same problem, in theory) or
    b) Eliminate the network altogether.

    Bush is asking for something that isn't possible because social engineering and the "inside job" is the oldest way to hack any system of anything. Hacking didn't start with computers, bank vaults, locks, jewelry stashes... they were all done in the past with inside work.

    It's impossible because of human error and human presence.
    • The only concievable way to do this is to either:

      a) Eliminate Government Data Access to All But the Highest Officials (which still poses the same problem, in theory) or
      b) Eliminate the network altogether.


      We already went down this path with the CIA and NSA. Turning to more hardware meant that we were less adapatable, and missed more things.

      While people will always be the weak link of any network, and inside access the way to defeat security, this does not mean that it is unwise to trust people.

      Instead, we should make security transparent and easy to use, and learn from our mistakes.

      This is the lesson of open source - the security actually increases as the number of eyes peering at the code increases. Dependence on the technology ignores the fact that someone has to see the data at the beginning and end of the process.

      -
    • Of course you are right, but you're missing the point somewhat. Of course no useful system can be totally secure. However just because the system isn't perfectly secure doesn't mean we shouldn't have any security measures in place. The fewer points of vulnerability, the easier it is to control and monitor those parts of the system that you can't secure technically.

      What Bush wants is not "poison-free food" but to make sure that the more egregious security problems of the Internet are solved. To extend your metaphor: if the ovens are unlocked, the food is never tested and the staff can't be trusted you're pretty much guaranteed a less-than-poison-free Thanksgiving feast.
    • That's why there are such things as security clearances, background checks, access lists, security officers, etc.

      No system is perfect. That doesn't mean that it isn't worth it to build a secure network. A security officer once told me that any system could be cracked, it was just a question of time and resources. The art of security is to make the cost of breaking into the system higher than the value of the information being protected. He said that the government had tested all of our locks and safes, and knew how long it would take an expert to crack them. They didn't have to be perfect, just good enough to stall an attacker for a specified amount of time.

  • Mae West/East (Score:3, Interesting)

    by lrc ( 5755 ) on Monday November 19, 2001 @06:41PM (#2586841) Homepage
    I've been wondering just how susceptible Mae West and it's ilk are to terrorist attacks.

    It seems to me that it wouldn't take a whole lot of bang to bring the internet to it's knees.

    Funny how it was originally designed to be immune to this sort of stuff.
  • by ez76 ( 322080 ) <slashdot@@@e76...us> on Monday November 19, 2001 @06:42PM (#2586848) Homepage
    Perhaps in the spirit of bipartisan cooperation, he could contract Al Gore to invent one?
  • Bush administration is considering the creation of a secure new government communications network separate from the Internet that would be less vulnerable to attack and efforts to disrupt critical federal activities.

    That's funny, I've always wanted the creation of an insecure anonymous non-government communications network separate (or on top of) the Internet that would be less vulnerable to efforts to regulate non-critical non-federal activities.

  • by pdqlamb ( 10952 ) on Monday November 19, 2001 @06:42PM (#2586854)
    None of the major backbones are willing to provide IPv6 connections. The U.S. Government contracts out almost all of its long-haul communication requirements. They used to get AT&T to build underground bunkers for them, but now they get nothing. Why not start by requiring IPv6 in all government RFPs/RFQs for long-haul comm? That should provide an instant market to kick-start IPv6, complete with all the security features that have already been designed.
    • None of the major backbones are willing to provide IPv6 connections.
      Bullshit. None of the major backbones are willing to provide IPv6 routing because IPv6 is still experimental for the next several quarters, and I assure you they're as desperate for a gimmick as the rest of the technology sector, or more so. If you think it's so damn easy, buy a Cadence or Synopsys license, take the risk, and do it already.
      Why not start by requiring IPv6 in all government RFPs/RFQs for long-haul comm?
      What does IPv6 use for security? It uses IPsec encapsulation and authentication, exactly the same as IPv4 save that it's not optional in IPv6. What's the advantage? We don't even have an address assignment scheme for IPv6 yet that's known to scale, and IPv6 users and early adopters need to work the bugs out as the scale of the system grows. Do you want routers to die or run impaired just because some non-conforming implementation tries to send a packet formed just wrong? Neither do I, and good infosec does things correctly, not quickly.

      There are ZERO operational advantages to carrying classified information over the public network when you are an organization of this size. You get a lack of control over the availability and of the network as a whole, and a nonzero possibility of leaked information via covert channels. Strictly divorcing the government operations network, properly done and with appropriate physical security applied to end-user terminals, reduces the chance of information leakage to zero and gives the network operator absolute control over availability, reliability, and access.

      If it were such a bad idea, then why do so many large corporations lease lines between offices?

      -jhp

  • Gresham's Law (Score:2, Insightful)

    by sharp-bang ( 311928 )
    I'd be really interested to know how Mr. Clarke et al are going to come up with believable cost figures for this unhackable network, particularly as what makes a network hackable is NOT so much the routers, bandwidth, etc. as the due diligence done by the managers, which is an ongoing expense. (The exception might be for a physically secure signalling infrastructure... anyone know how to keep a physical network from being blown up or jammed?) But I just don't see how this would hold up in the long run... bad security inevitably drives out good if human operators (and usability drivers) have anything to do with its maintenance. Perhaps the money would indeed be better spent deploying IPv6 on a large scale, which is probably the only way we will see it replace IPv4. Since this network ultimately subsume the existing Internet or be subsumed by it, it seems best to keep this end in mind.
  • It might be a better idea to support research into strong encryption, good protocols, etc. Maybe. But this is a pretty good idea. Think of all the boneheaded things they could have done instead: outlawed tools that could potentially break encryption. Outlawed computers that don't pass a "security audit" which required that all security-related source code be closed (effectively killing off Linux). Or worse still, done nothing and left sensitive government data floating around on the Internet, weakly encrypted.

    This isn't a half-bad idea. A private network is still of course vulnerable, but it's like putting a fence around your property. People might still end up on your property, but they'd have a lot harder time explaining why they're there, rather than just "uh, I just got lost".

  • by The Dev ( 19322 ) on Monday November 19, 2001 @06:50PM (#2586903)
    If the current telco and internet infrastructure is any example, their efforts will do no good. A dozen terrorists with rented (or commandeered) backhoes in select locations could cause massive disruptions in the Internet (and therefore the economy). Miss Utility could even be an unwitting accomplice.

    Don't even start with "physical diversity blah blah blah". The fact that your physically diverse circuits aren't has been proven time and again by the mighty backhoe/flaming hazmat car/junior achiever.

    Of course some improvements to BGP wouldn't hurt either.
    • The fact that your physically diverse circuits aren't has been proven time and again by the mighty backhoe
      And even the mighty backhoe takes doing to impede the satellite or the carrier pigeon. If you've got such a large organization, and the data Absolutely Positively Has To Be There and Absolutely Positively Has To Remain Private, you use diverse media and serious encryption.

      Trust no one, not even a sweetheart government contractor.

      -jhp

  • by j7953 ( 457666 ) on Monday November 19, 2001 @06:50PM (#2586905)
    Isn't hard-to-disrupt communication the reason that DARPA got involved in this "Internet" business anyhow?

    Yes. And the internet itself is hard-to-disrupt.

    However, a single server can be the target of an attack, and this is what they want to secure against now. The idea of the internet was to be able to communicate even if lots of nodes failed (i.e. got physically destroyed). The idea was not to secure every single node against destruction. Also note that the internet was designed with physical rather than digital attacks in mind.

    The government certainly does have a point here, but I think you can reach security for each individual node only by securing those nodes, not by simply seperating them. How will they make sure that, for example, no email can get in from the internet? Have two computers at each user's desk?

  • AUTODIN (Score:3, Informative)

    by pete-classic ( 75983 ) <hutnick@gmail.com> on Monday November 19, 2001 @06:51PM (#2586911) Homepage Journal
    AFAIK AUTODIN is still where the "serious business" happens.

    AUTODIN is an ancient, circuit switched network. It's a real bear to operate (I spent four years operating it) but it is genuinely secure. AFAIK the whole "packet switched so it can't be decapitated" thing that the APRANET was supposed to solve was supposed to be an answer to AUTODIN.

    I hope they get something going so they can retire AUTODIN.

    -Peter
  • The hosts on it are also important. Now most people don't want to use overly secure systems (B2 level can become quite painful, but is actually required to prevent users from executing arbitrary code received over the network), so host security will remain low. Even if you separate the network from the other internets, one security breach can still have devastating results. And since people tend to keep modems in their drawer in order to log in from home, security breaches are going to happen.
  • by weave ( 48069 ) on Monday November 19, 2001 @07:01PM (#2586953) Journal
    All it takes is one idiot to install PCAnywhere and throw a dialup modem on their office computer so they can work from home. Or someone who dials out to the net from their office computer and runs something like Go to my PC [gotomypc.com].
  • Al Gore (Score:3, Funny)

    by Tom7 ( 102298 ) on Monday November 19, 2001 @07:22PM (#2587025) Homepage Journal

    Somehow this whole discussion would be a lot funnier if it was Al Gore saying that he wanted his own private internet.
  • Given their cozy relationship they'll probably want to use Microsoft's latest server [bbspot.com] which is the only one proven unhackable.
  • Simple

    George Busth will never forgive the internet for allowing itself to be invented by Al Gore.

    So he is going to redo the whole things and invent the BushNet, a secure unhackable network based on the ingenious idea of running the following script on all government machine:

    #!/bin/sh

    rm /dev/eth0
    ln -s /dev/null /dev/eth0

  • Or is there something else a 'net terrorist' could do to 'disrupt the vital flow of information'?"

    I thought this was the government's job, not the terrorist's job.
  • My initial impression is that the net would be less prone to complete shutdown than other infastructure. The net still is sort of a wild wild west, and everybody from skript kiddies to hackers are continually trying to break in and DOS various different sections of the Internet. It's hard to imagine how any group (unless it was some massive government funded operation) could be more disruptive than what currently takes place. Radical islamic fundamentalists dont' seem THAT tech savvy.

    Airports thought about security a bit, but really serious measures generally weren't taken. However, security has been one of THE TOP issues for the Internet for a long time. Kerberos, ssh, bastille linux etc... there are a lot of tools out there to lock systems and networks down.

    That said the government is probably getting hacked all the time now. Really critical systems probably should physically seperated from the net. One aspect of security that is the most difficult is human error. Sure a system can provide ssh and kerberized login, but if people use the same password for their yahoo games account, all the encryption in the world doesn't appear to do a lot of good.

    Just some random musings.
  • by sterno ( 16320 ) on Monday November 19, 2001 @07:32PM (#2587067) Homepage
    The notion of a secure private network for the government seems like a decent idea. To think that through such a private network we can avoid some sort of internet peral harbor is absurd. Why? Real simple: was the world trade center a government building?

    Why would any terrorist waste their time and resources trying to take down the FBI when it could go after banks, airports, power grids, and a whole host of other things that are on the public Internet? All of those things are far more visible and have a far more significant immediate impact on the lives of US citizens. Remember, terrorism isn't about taking out strategic assets, but creating a sense of fear in the every day lives of normal unassuming people.

    Now, one might say that the answer to this quandry is to put corporations on that network. Of course then you are expanding the base of users and increasing the likelyhood that a few terrorists (or those easily bribed or fooled by them) will be able to breach that network. I suspect that even putting large swaths of the government on that network already risks that compromise within the government itself but that just amplifies it.

    Why don't we take that money and put it into developing policies and technologies that will make the current networks more secure? I know that this doesn't look as impressive to the public, but in the long run it will probably do more to prevent an Internet Perl Harbor.
  • Physical security (Score:3, Insightful)

    by cr@ckwhore ( 165454 ) on Monday November 19, 2001 @07:38PM (#2587085) Homepage
    Building a private network isn't a big deal. I think the government could build an encrypted WAN without much effort. I think the biggest challenge to security is going to be on the physical front... meaning that every piece of network equipment must be in a secure location. This includes every router and bridge in every network shack along the WAN lines. Wouldn't want any 1337 hax0r5 to come along with a patch cable and bring down the government network. Since guarding every inch of wire is impossible, point to point connections must be made with fiber line so it can't be tapped like copper.

    None of this even begins to consider the physical local machine security... government workers shouldn't be alowed to bring any media from home, no incoming modem lines, etc.
    Lots to think about. If GB wants to cut me a check, I'll begin the engineering work tomorrow.
  • Turning to other news tonight, new reports on the status of Unhack-a-Net, originally proposed by former President Bush, indicate the test servers were actually transmitting gps information to would-be hackers, indicating their course and heading.

    And in an ironic turn of events, an undisclosed number of people were arrested in nationwide raids following the most recent round of Unhack-a-Net testing, on charges of using illegal circumvention devices. Officials close to the case described the devices as 'Garmin eTrexes.' The official hinted at prosecution under the SSCA (Super-Secret Copyright Act), the details of which are still classified.

    One detainee was overheard saying, "But...we're beta testers! You know, Unhack-a-Net!"

    SSCA was signed into law in 2003, following the terrorist threats to the music and film industry. Those attacks came in the form of the thirteen year-old son of a record company exectuve, who crashed his father's Windows 2000 computer one night. Under the terms of the MASTA (Microsoft Antihacking, Security, and Terror Act), the child was sentenced to a prison term, but President Ashcroft felt greater protection was needed for America's vital interests.
  • Uhh, milnet? (Score:2, Informative)

    by Omega ( 1602 )
    ...the creation of a secure new government communications network separate from the Internet that would be less vulnerable to attack and efforts to disrupt critical federal activities.

    Doesn't MILnet do this already? Isn't this why when the DoD gave up control of ARPAnet, they forked and created MILnet to retain a secure channel?

    Bush needs to lay off the MSN. The U.S. government is already waaaaaaaaaay ahead on this one.

  • by peter303 ( 12292 ) on Monday November 19, 2001 @07:54PM (#2587134)
    The problem is that open networks evolve so much faster than closed, secure networks, that users become frustrated with the later and start moving files surrepticiously between them. Thats what Prof Deutch of MIT did while head of the CIA and Wenho Lee of Los Alamos.
    • Thats what Prof Deutch of MIT did while head of the CIA and Wenho Lee of Los Alamos.
      It can be made impossible (read: "prohibitively difficult") for most people to move data off of the red network without infosec officers noticing. simply by defining your network border to include end-user terminals and securing the network to match. Yank the floppy drives, lock down MAC addresses on switch ports, ban CD writers, install tamper switches in the cases. Ban cameras, save copies (hard or soft) of everything that gets printed, control physical access to printers, embed radio security tags into the paper. A rogue user can always lie about why they're removing plaintext classified information from a classified network, but if they can't get it off the network, they can't get it out of the building.

      As for open vs. closed networks, who cares about evolution? If you've got the tools to do your job correctly, you don't need anymore.

      -jhp

  • by catseye_95051 ( 102231 ) on Monday November 19, 2001 @07:55PM (#2587139)
    We alreayd have such a network. Its called milnet and is used by the US millitary who funded the original inetrnet research.

    As soon as the internet was working they built their own, secure network, and got the hell off of the publicly acessible one.

    Maybe Colin won't let Georgie play with his toys, so Georgie wants his own....
  • That the US Govt saying they want to do this is akin to a company saying they want to build a large, private WAN, because they don't like working on the internet for sharing info between offices. Fair enough.

    Apples and Oranges.
  • by rice_burners_suck ( 243660 ) on Monday November 19, 2001 @08:07PM (#2587180)

    Even with a private network that isn't connected to the Internet, there is still at least one big security issue: A false sense of security. Government employees may think that because their private network is so secure and separate from the big bad Internet, they can relax and give computer security a low priority. What most folks don't understand is that computers are like any machine: They require constant maintainence for reliable operation. Security is a large part of that maintainence, and cannot be set aside while other things take place. On the contrary, security must proactively be part of everything that goes on in a computer and network. This is partly why a false sense of security is dangerous.

    Besides, intruders could still access the network through such techniques as war-dialing, to name one example off the top of my head.



  • Whats Bush Talking about? The government has had independent secure private internets since before we even had the internet.

    Why are they telling us what they are building unless its going to be a public government internet.

    I mean really, if something is private and secure, the last thing to do is tell the world about it.

    When the government wants to keep secrets they can, and they do so by not telling us anything about it,

    Perhaps bush wants an internet seperate of the private government internets already in place so he can email his friends in various other countries on any computer (not just the secure private ones) without worrying about people reading his msgs.
  • by Apreche ( 239272 )
    first of all nothing is unhackable. Second they're talking about setting up a seperate wan for just the government. If just ONE computer on that network is also connected to the real internet, then someone can get in. If none of the computers on that network are connected to the internet, then government employees will be very unhappy at work. Hence, another waste of money.
  • Some basic things can be done to make "secure" or "segregated," or other types of somewhat-more-protected-than-usual environments.

    Unfortunately, I think that there are also some very real problems. Some very old military systems (e.g.) SAGE - were secure. The customer (Government) could own and have all code reviewed. All end points were well controlled. The number of nodes and links, etc... were limited. The system was also special, and dedicated - purpose.

    There are limits as to how secure any system will be if it will be built on off-the-shelf components, software and hardware components that the gov't can't fully inspect, networking protocols that are not provably secure, and the inevitable ... using currently available products to implement solutions, rather than building that which might be necessary.

    Sam Nitzberg
    sam@iamsam.com
    http://www.iamsam.com
  • If the government wants a really secure network of nontrivial size, then it probably should not use TCP/IP as its underlying protocol suite. TCP/IP was designed in the 1970s for a limited-access insecure network of researchers (ARPAnet). If anyone misbehaved, they'd be booted, and/or their site manager would get a nasty notice. Nobody was "entitled" to be on ARPAnet, and almost everyone cooperated. The network was designed for maximum openness within that selected community.

    Now we have the public Internet, and Microsoft's virusware for applications. Firewalls help, but as many have noted, it's too easy for a laptop or floppy to inject something, and if an email gateway it provided, MSware will do the rest. Or any other mail client that follows their evil lead and executes email.

    A serious fix is to create a new protocol suite that has security designed in. New stack code with no buffer overflows. A stack that doesn't invite address spoofing, flooding, or various other vulnerabilities of TCP/IP. Not that TCP/IP is all that bad for public use, but you just don't try to add security later and expect it to work! (It's a sieve: It should stand for Transmission Colander Protocol/Insecure Protocol.)

    This new stack would have new, or at least modified, applications written for it, the way ARPAnet did back when it was young. And rules against insecure crap, so no Outlook ports! It might then catch on outside, but if the protocols have security handles in them, it's okay; there's no security through obscurity. This would help long-term stabilization of the public Internet, if it adopted more secure (and probably more efficient) protocols. Just as government funding for its own use led to TCP/IP.

    Some people seem to think that TCP/IP was handed down to Moses on Sinai, and is thus sacred, Perfect, and should be inviolate. I don't buy that for a minute, and I was on the ARPAnet back in the NCP days. It was a nice experiment but it has ossified with widespread use, and clearly has trouble keeping up with current needs. IPv6 is not an improvement in any sense, efficiency or security; it is a distraction whose misbegotten presence, on balance, makes things worse.
  • Damn it (Score:3, Interesting)

    by dimator ( 71399 ) on Monday November 19, 2001 @10:57PM (#2587652) Homepage Journal
    I love it how the /. editors always have an excuse as to why they post dupes. Either it's witty, or dodgy, or it's "this is important enough to read twice." Please.

    Is it THAT IMPOSSIBLY HARD to use your OWN search tool before posting dupes?
    • A cat that comes when you call it.
    • An oven that doesn't burn things.
    • A silent chainsaw.
    • Enough RAM.
    • Wishing doesn't make it so, Mr. President. Networks are designed to let people share information. Even if you cut yourself entirely off from the Internet, you leave yourself wide open to moles, leaks, and all sorts of human error. A private network may make the human security holes even wider because it gives you a false sense of safety. I'd rather see my tax dollars spent on secure open protocols and sensible security policies. Security is a mindset, not a technology.

  • Right now this thread is filling with posts about why or why not this network will be secure, and why or why not all of the OTHER protected/secret government networks are/are not secure. What people are missing out on is that the government does not actually WANT a secure network.

    Bush and co. want a new network because two states, California and Viriginia, are full of out-of-work techies, left jobless by the dotcom collapse. Virginia and California are also the top two states in regards to defense agencies, contracts, locations, dollars, etc.. Building a new government network would create a huge number of stable, high-paying jobs in Virginia and California as the agencies and contractors in those states were wired up; and even more jobs all across the country as the network spread out to all of the other states in between.

    Not only does this have the effect of greatly boosting the economy without pissing too many people off (Which Congress has proven they cannot manage to do.), it also earns a lot of loyalty to the Republican party from all of the people who get those jobs, as well as the other people who benefit from those jobs as the money trickles outward.

    Is this network needed, or even likely to work? I do not really know, and anyone who had nothing better to do than post to Slashdot about it really does either. But that does not matter, because right now America's economy needs to get going, the world needs our economy to get going, and the people making decisions in the White House realize that this is a good way to give a long term boost to the economy and their careers, without really earning much scorn, and they would be fools not to.

"When the going gets tough, the tough get empirical." -- Jon Carroll

Working...