back to article ‘Mother of Internet’ Radia Perlman argues for centralized infrastructure

Internet pioneer Radia Perlman has argued in favor of centralized infrastructure, while speaking at the International Symposium on Blockchain Advancements in Singapore on Friday. Perlman said that conventional wisdom poses centralised systems as “bad” and decentralized as “good,” however decentralized entities are inherently …

  1. jmch Silver badge

    Horses for courses

    In most cases, a centralised infrastructure is what you need because that's the one that can best scale, operate more efficiently etc. By "centralised" I don't necessarily mean there is one channel / processor, but that there are multiple levels of grouping, exactly like e tree from leaves to branches to bigger branches to trunk (and ideally having multiple trunks). Of course it requires trust in that 'trunk', but you also get legal oversight and ultimate responsibility.

    The blockchain is completely peer-to-peer with every peer communicating with every other peer. This is terrible for scaling (which is why Bitcoin can only process a handful of transactions per second), and it's *only* advantage is that it is completely "trustless" in the sense that you do not have to trust a central controlling authority. But as Perlman points out, if you're transacting with an individual, you still have to trust that they will deliver what you are paying them for. Of course if I use bitcoin to order something online from a reputable company and it doesn't get delivered I can still sue the company just as if I paid by credit card, as long as there is a transaction record such as an email order confirmation - there is a public record of the actual transaction on the ledger. It's for illegal stuff like buying drugs or ordering a hitman that someone doesn't have any recourse if they're defrauded.

    I also see here a very skewed view of Bitcoin - it's used to 'hide' from government, therefore is used mainly for criminal purposes. That reflects a complete misunderstanding of why Bitcoin is mostly used - as a store of value, not as a transactional currency. As many silk road drug dealers found out, Bitcoin is not, in fact, anonymous. It is at best pseudonymous, but real wallet owners can be identified through transfers within the Bitcoin network, and between Bitcoin network and fiat accounts. The main 'selling point' of Bitcoin is that no single entity can "cheat" by creating new Bitcoins without proof-of-work, the amount of Bitcoins mined is limited, and the ratio of newly-mined coins to existing coins is tiny (thus giving it value-holding characteristics much more similar to those of gold). On the other hand, value stored in fiat currency as cash or savings are eroded by inflation and cost-of-living increases, and "safe" investments give returns that are worse than inflation.

    1. Crypto Monad Silver badge

      Re: Horses for courses

      The blockchain is completely peer-to-peer with every peer communicating with every other peer. This is terrible for scaling (which is why Bitcoin can only process a handful of transactions per second)

      That's only *one* reason why Bitcoin is terrible at scaling. There are other consensus protocols, like Raft, which scale pretty well for their use cases. It's the proof-of-work part of Bitcoin that bites, i.e. you have to burn real physical resources to be a significant part of the blockchain.

      it's *only* advantage is that it is completely "trustless" in the sense that you do not have to trust a central controlling authority

      Instead, you have to trust a cabal of mostly-anonymous shady organizations who are the big Bitcoin miners. If they collude, they *can* reverse a blockchain transaction. It doesn't happen very often, but it does - like when 184 billion bitcoins were created out of thin air by an integer overflow bug, and they decided that wasn't a good thing.

      But they decide based on whether it's a good thing *for them*, not for anyone else.

    2. JoeCool Silver badge

      Re: Horses for courses

      "... therefore is used mainly for criminal purposes"

      That was not said. I think you are making an inaccurate implication. She is pointing out the "Unique Sales Proposition" of Bitcoin. You can argue on that point, if you want to distract from her overearching points.

    3. vtcodger Silver badge

      Re: Horses for courses

      Bitcoin is mostly used - as a store of value

      The current bitcoin price is $16,956.30. The all time high value on Nov 8 2021 was $68,789.63. To get back to the Nov 2021 value, Bitcoin needs to increase by 305%. Not my idea of a store of value

      Gold on the other hand .. Nov 8 2021 = 1,824.58. Yesterday 1,803.30. Increase required 1%. I think the "Store of Value" argument for Gold has some merit. Bitcoin. Not so much.

      I'm not especially a fan of Gold for reasons too lengthy to go into here. But compared to Cryptocurrency there is a LOT to be said for Gold and other traditional value stores.

  2. oiseau
    Facepalm

    Some common sense

    Friday at last.

    ... described AI algorithms that lead to polarizing rabbit holes of content as among the top dystopian characteristics of the internet.

    ... allows disaffected extremists to connect with one another.

    Yes, reads/sounds like common sense.

    But that is only one aspect of the harm AI can unleash on civilization.

    It can also allow a great many more dangerous things to happen/be done. eg: armed dog robots anyone?

    ... usually means that the data is stored in lots of places, so your data is not going to get lost.

    I beg to differ.

    This would happen only if you sufficiently stress the adverb usually.

    And if the phrase "lots of places" includes your own mandatory/secure on-site mirror.

    ... especially if your data is stored in a public cloud.

    Exactly my point.

    "... its clear who to blame when things go wrong ..."

    "... someone to answer for the system’s problems."

    Right.

    No need to explain or cite the many past examples to understand that this is not so.

    Once the Shit Hits The Fan ® what you want is your bloody data back, not ASAP but right now.

    Not to know the name of the dickhead that let things slip or the beancounter at the cloud facility who, in the name of shareholder dividends, decided to cut corners and caused the problem, among many other possible/usual causes or culprits.

    Because once your valuable data went south, it is gone.

    How much is that worth?

    O.

  3. that one in the corner Silver badge

    “I think pianos are wonderful, but I wouldn't use them for mass transportation."

    Wonderful phrasing.

    Really hope to see to that used when more inappropriate blockchain usage pops up.

    Or any other daft wrong-tech-wrong-place.

  4. Howard Sway Silver badge

    I'm not sure what you would buy with Bitcoin

    Well, all the self-driving piano startups only accept payment in crypto.

  5. Anonymous Coward
    Anonymous Coward

    Perl is confusing

    > “Centralized means one organization is in charge. It usually does mean that there are multiple servers so it doesn’t mean single point of failure. And usually means that the data is stored in lots of places, so your data is not going to get lost. And especially if your data is stored in a public cloud,” said Perlman.

    Perlman seems to talk solely of the technical "single point of failure", and ignore the business and political ones. Just look at Twitter for how a centralized system can be _technically_ working perfectly, but be on the verge of collapse politically and financially - all because of its single centralized authority.

    Politics is the entire reason for Bitcoin and blockchain to have been invented. There are indeed a myriad of more technologically "efficient" ways of transferring money or value than Bitcoin, but they are all subject to the whims of a dictatorial political authority, usually beyond your control or oversight. On the other hand, the only authority behind Bitcoin is the consensus of those contributing to the Bitcoin network.

    > “Now if you're using Bitcoin I'm not sure what you would buy with Bitcoin, probably something like a hitman. And if he doesn't kill, who could you complain to? How do you get your money back? So most of the time, centralized is exactly what you want,” she further reasoned.

    This is a nonsensical argument to use. If I hire a hitman with cash, and they don't do it, who could I complain to? It would be ridiculous to suggest one should complain to the national mint to get a refund for a broken contract! No, you go to the place which handles contract disputes - the courts (or the mob boss, etc.). This applies no matter what mechanism you use to transfer value - cash, gold, Bitcoin, shares, pork bellies, magic beans.

    1. Rafael #872397
      Joke

      Re: Perl is confusing

      I'd say. I still cannot understand why is there a "unless" statement, and why "use strict" is not a default feature :-)

    2. Charlie Clark Silver badge
      Stop

      Re: Perl is confusing

      Perlman seems to talk solely of the technical "single point of failure", and ignore the business and political ones.

      Nope, you seem to be cherry picking for the sake of your own argument. She's referring specifically to governance, ie. business and political aspects.

      Just look at Twitter for…

      This is a terrible analogy. Twitter isn't any kind of technology but an advertising company based around the inane things people say. A better analogy would be, oh I don't know, the internet DNS.

      1. Anonymous Coward
        Anonymous Coward

        Re: Perl is confusing

        > Nope, you seem to be cherry picking for the sake of your own argument. She's referring specifically to governance, ie. business and political aspects.

        I quoted an entire paragraph from the article, where she clearly is quoted as saying: "It usually does mean that there are multiple servers so it doesn’t mean single point of failure". "Servers" are technology, and have nothing to do with governance.

        Where in the article does she address the single point of business or pollitical failure that "Centralized means one organization is in charge." brings?

        > Twitter isn't any kind of technology but an advertising company based around the inane things people say.

        What a nonsense argument. You think Twitter doesn't use a LOT of technology to provide their advertising and facilitate people saying inane things at the scale they do?

        Certainly they have "multiple servers" and "data stored in lots of places", so there's no single point of failure right? No problem with a single organization being in control of it all, right? No chance that a change in management may bring the whole thing to ruin, right?

        > A better analogy would be, oh I don't know, the internet DNS.

        I wasn't making an analogy - I was giving a direct example of the business and political risks of centralization of services.

        Have you taken your dried frog pills today?

        1. Charlie Clark Silver badge

          Re: Perl is confusing

          Thanks for the reductivist ad hominem nonsense!

    3. Peter2 Silver badge

      Re: Perl is confusing

      This is a nonsensical argument to use. If I hire a hitman with cash, and they don't do it, who could I complain to?

      The police in most countries, or a court. Accepting the money in exchange for doing something does form a contract under english law, and failing to perform it is breach of the implied contract terms and all.

      https://www.independent.ie/world-news/europe/woman-gets-2000-after-hired-hitman-failed-to-kill-her-26404390.html

      I can imagine the judge having trouble keeping a straight face dealing with that one, since the only defence possible against failing to perform the contract is to demonstrate that you did try and perform it, which amounts to a confession to conspiracy to commit murder on record in court, which can be used to prosecute you for that offence.

      1. doublelayer Silver badge

        Re: Perl is confusing

        This is true, but it works just as well with Bitcoin as with anything else. If it's a hitman I'm hiring, then I'm not likely to go to court and by my accusation prove culpability for a crime, but let's consider something legal. My domain registrar, Gandi.net, has a Bitcoin pay option. Let's say that for some reason I decided to abandon my simple credit card payment for whatever morass is involved to send them enough Bitcoin that it turns into the right amount of money when they're done converting it. If I send them Bitcoin and I don't get a domain name, I can take them to court and would send complaints to that effect. This works whether I paid them in Bitcoin, with a payment card, by mailing cash, or any other method. The fact that Bitcoin is decentralized and credit cards are not makes no difference, given that either way I'll be using a centralized court to resolve the contract dispute.

      2. Timochka

        Re: Perl is confusing

        *Adjusts glasses*. Well, AKKshually...

        The courts wouldn't help you in this case, because for the specific case of the hitman, no contract actually exists. It is one of the basic tenets of contract law that a contract to achieve an illegal purpose (e.g. murder) is in and of itself illegal, and thus essentially cannot exist in the law.

        Disclaimer: Dredging up memory of contract law lectures about 30 years ago. Also, England & Wales law, other jurisdictions may vary.

  6. b0llchit Silver badge
    Mushroom

    Broken? No thanks.

    So aren't you grateful to us that we gave you such a broken thing?

    No, I'm not grateful. I'd rather have you'd actually solved problems than to unleash them onto the next generation. Then we could stop worrying about the legacy and invent nice things.

    1. Claptrap314 Silver badge

      Re: Broken? No thanks.

      When our ancestors solved the "the lions are killing us" problem, it was immediately replaced with "our neighbor lion-killers are killing us". What is termed "progress" has always had this tendency to replace one problem with another.

      There is a good reason to focus on today's problems. Most of the developments of the sort she seems to be mentioning were the secondary effects of earlier problems being solved.

    2. jake Silver badge

      Re: Broken? No thanks.

      As I've been saying in this forum for years, the Internet is nothing more than an insecure research platform gone mad.

      When we built this thing, we knew there were issues, even pre-BARRNet and NSFnet. It was not ready for prime-time back then, and it's still not ready for prime-time. In fact, it'll never be ready for prime-time, no matter how many bandaids we apply. It'll remain an insecure research platform, at least until it is redesigned from the ground up. At which point it will no longer be The Internet.

      We tried to keep the corporate world out of it because of this, but we were naive. We failed. We apologize.

      I'm 99.9% certain that I can speak for us on this matter, especially seeing as TINU.

      1. doublelayer Silver badge

        Re: Broken? No thanks.

        "I'm 99.9% certain that I can speak for us on this matter, especially seeing as TINU."

        Well, apart from my not having a clue what "TINU" means (I doubt it's Tubulointerstitial Nephritis, and the only relevant acronym I could find online was "totally incompetent, not useful" which doesn't really fit either), I don't consider you to have that right. You're a pseudonymous person online. I only have your word for it that you were around for early experiments with the internet, and since I don't know your identity, I also don't know what your role was. If you had a minor part which allowed you to see the early experiment, then I wouldn't assume you know the views of all the groundbreakers. Let's assume that you had a central part. I still wouldn't assume your opinion is held by everyone else who could have expressed such a view a lot more strongly had they wanted to.

        If you really did design the internet, thank you for it. It's not perfect and it never will be, but I doubt a redesign would really fix the issues that we face with it anyway, since many of those are political or economic rather than technical. Like all of human invention, it will be a mess of different standards, attached together with tape, with bits falling off from time to time, and still providing more by its chaotic availability than it would being stuck in a quest for unachievable perfection. Although if you have an internet 2.0 plan lying around, maybe we can consider spinning up an experiment of that one for a possible replacement or at least some cross-pollination.

        1. jake Silver badge
          Pint

          Re: Broken? No thanks.

          "Well, apart from my not having a clue what "TINU" means"

          "There Is No Us", a tongue-in-cheek reference to the Backbone Cabal, which pushed through the Great Renaming of Usenet, and generally kept network abuse[0] to a background murmur in the 1980s. The folks involved publicly denied their own existance in order to cut down on conspiracy theory bullshit clogging up the pipes. Most cognizant users (including the Cabal membership) would sign their Usenet posts on the subject with "There Is No Cabal". It's an odd fact of history that the TINC acronym came into wide-spread use after the Cabal split up in roughly 1988 (the mailing list stopped at that point, but most of the regulars maintained contact for technical matters for decades after. Still do, in most cases. TINC.).

          "If you really did design the internet"

          Hold up there, pardner. I said no such thing. I have never stated that I designed The Internet. However, I was an engineering student at Berkeley and Stanford from the mid '70s through the '80s, and did contribute. As did many other people.

          "Although if you have an internet 2.0 plan lying around"

          We are using it as I type. Internet 2 went live on January 1st, 1983, when we switched from NCP to TCP/IP. I was discussing a redesign from a technical perspective would be necessary for a non-research, secure network. The Human part of the equation is out of our reach, thankfully. THAT kind of engineering I'm not interested in.

          Beer?

          [0] That's abuse OF the 'net, not abuse ON the 'net, if you get the distinction.

  7. mark l 2 Silver badge

    “Now if you're using Bitcoin I'm not sure what you would buy with Bitcoin, probably something like a hitman. And if he doesn't kill, who could you complain to? How do you get your money back? So most of the time, centralized is exactly what you want,”

    I remember seeing some documentary where they basically said hiring hitmen on the dark web is ALWAYS a scam. When the police investigate where a murder took place and a hitman was paid to kill somebody its usually found out that the hitman and the person taking out the hit knew each other in real life, and that killer isn't a professional hitman but is usually someone desperate for money and of low intelligence. Unless of course you are already involved with organised crime such as the drug trade etc and know the right people to approach and those people don't advertise on forums.

    But maybe Radia is saying we need a centralized organisation for hiring hitmen where we can pay with Mastercard and VISA? LOL

    1. ExampleOne

      So, basically, an Assassins Guild?

  8. nijam Silver badge

    > ... And if he doesn't kill, who could you complain to? How do you get your money back? So most of the time, centralized is exactly what you want...

    Which centralized authority for hitmen does she use, out of interest?

    1. Throatwarbler Mangrove Silver badge
      Holmes

      The CIA. Duh.

      1. Anonymous Coward
        Anonymous Coward

        "The CIA. Duh."

        I thought so too, but then I remembered, that was RON Perlman.

  9. Andy 73 Silver badge

    So this is a political site now?

    Compare this article - in which the subject is critical of blockchain technology whilst talking at the Singapore Symposium on Blockchain, with the article by the same author in which another subject is critical of blockchain technology whilst talking at the Singapore Symposium on Blockchain.

    It's pretty pointless coming here for technology coverage if all we get to hear is the author's personal politics.

    1. doublelayer Silver badge

      Re: So this is a political site now?

      Or alternatively, the author concerned was assigned to cover the Singapore Symposium on Blockchain and the main news they thought would interest the reader was the statements being made there by the speakers invited there. Since when did covering someone's statements about a specific technology become political, and why do two articles from a symposium somehow eclipse the many non-blockchain articles you can find on the home page?

      1. Andy 73 Silver badge

        Re: So this is a political site now?

        I suggest you follow the instructions at the start of my post and actually *compare* the two articles. The "mother of the internet" is given favourable coverage for saying that blockchain ain't necessarily that useful. The ex-Tory leader is roasted for saying that blockchain ain't necessarily that useful.

        1. jake Silver badge

          Re: So this is a political site now?

          You're wearing it wrong. Shiny side OUT.

    2. Irony Deficient

      Re: So this is a political site now?

      Two articles by one author presenting one point of view at one symposium does not a political site make.

      If you don’t like an author’s point of view on a particular topic to the point of declaring a site’s technology coverage “pretty pointless”, then perhaps not reading that author’s articles on that site could be the basis of a simpler coping mechanism.

      1. doublelayer Silver badge

        Re: So this is a political site now?

        Not only that, but they're attributing the views of the speakers to the views of the author who covered them. Two people spoke in opposition to blockchain in some ways, ways that I note aren't entirely compatible with one another, and therefore they think the author must agree. They have weird ideas of politics and journalism.

  10. Anonymous Coward
    Anonymous Coward

    "Mother of Internet"...Really?

    She wrote the absolute best book on the subject of routers etc back in the 1990's. Everyone who does any i.p level network programming should read it. Even if its just WinSocks. Its a fantastically informative read. A must read. But as someone who has written networking stacks all the way from Ethernet drivers all the way up to the very long RFC protocols its a long way from LAN protocols like STP to what was used to build out Arpanet. Yes, I go back that far. I dont think there is even a STP RFC. Might be wrong.

    So enough of this political motivated re-writing of history. She wrote a classic book in networking. She was a key part of the writing of some very important LAN protocols. Thats all. Not the "Mother of the Internet" blather.

    Almost as silly as claiming that some guy who was part of the team that shipped a failed game console that almost no one heard of (even at the time) "invented modern game consoles". Total bollocks. Now the 2600 that shipped 10 months later, thats a very different story.

    1. PRR Silver badge

      Re: "Mother of Internet"...Really?

      > Not the "Mother of the Internet" blather.

      She did a LOT of protocol and routing study. No, if she was a guy we would not say "Father of the Internet", someone else gets called that. (Two guys, Vince and Bob?) Saying "mother" here is just sexist. But does make a snappy article title. And after all, isn't that what the Register is for? Not permanent truth but little jolts of fact-like snark.

      And author Dobberstein was probably very young when the internet was birthing?

      And Radia Perlman: Don't Call Me the Mother of the Internet in The Atlantic.

    2. jake Silver badge

      Re: "Mother of Internet"...Really?

      Yes, it is bollocks.

      What she did was pretty much single-handedly make DEC networking work. Not the Internet, which had already been evolving for some time, and was working nicely, before she joined DEC in 1980. Basically, she made all the little LANs dotted around DEC's world-wide campus play nicely with each other. Not a mean feat, that. But it was all on a private network, not The Internet at large. That's part of the reason none of her early work is written up in RFCs, it was all proprietary DEC. (Yes, DEC sold this technology to other people who needed similar WAN capability with their LANs. But this was early '80s, the Internet wasn't exactly business friendly at the time.)

      That is not to say her STP (and many, many other things ... Check out her prolific RFCs from 2008 to 2018) didn't help out the Internet later on. Far from it. She's a godmother of modern networking, make no mistake, and someone I will always look up to.

      1. Anonymous Coward
        Anonymous Coward

        Re: "Mother of Internet"...Really?.. all hail DECnet..

        I knew about the DECnet stuff. As someone who lived for a time in the DEC PDP8 / PDP11 / VAX universe for a while in the 1970 and 1980's. That and DG land.

        I always used DECnet as an example of how to do it right. It just worked and everything made sense. But when it came to building out the modern LAN / private WAN universe its influence was less pervasive. I say this as someone who considers the buyout of DEC by Compaq as one of the greatest disasters in modern tech. Up there with Oracle buying Sun. Two fanatic tech companies destroyed by third rate companies run by nasty non entities.

        I think losing DEC and then Sun is the reason why there has been so little tech innovation in the non mobile world for the last two decades. They created so much of what we use today. One way or another. And had so many fantastically creative people working for them at the time.

    3. Bebu Silver badge

      Re: "Mother of Internet"...Really?

      "Interconnections: Bridges, Routers, Switches, and Internetworking Protocols" 2/e

      Radia Perlman

      Pearson Education 1999

      > "Everyone who does any i.p level network programming should read it."

      Also really interesting for the non-TCP/IP networking eg Decnet phase iv which I assume is now ancient history.

      I recall another coauthored that was to me just as lucid-

      "Network Security: Private Communication in a Public World"

      Charlie Kaufman, Radia Perlman, Mike Speciner

      Prentice-Hall 2002.

      (Might be updated by "Network Security" by same authors Addison-Wesley 2020?)

  11. P. Lee

    A symposium on decentralised ledgers resistant to government interference, held in Singapore.

    So... it wasn't held in Beijing or Ottawa, but I think we all knew where this was going.

    I think bitcoin's use is pretty limited too, but sometimes the difficult and inefficient things are worthwhile.

  12. fxkeh

    "if there are 50 terrorists in the country it's no big deal"

    Leaving aside if that many *is* a big deal, the real danger is easily connecting people who want terrorism with impressionable losers who can be radicalised into carrying out terrorist acts.

  13. Anonymous Coward
    Anonymous Coward

    Actually Bitcoin…

    …was intended a secure alternative to PayPal or putting a credit card form on your website.

    The idea was, buy some bitcoins from a central site, transfer to a random website, they convert back to cash.

    The concept of using bitcoin as an investment was probably not intended. I don’t think it was even intended to be used outside an automated process involving “real” money…

    1. Claptrap314 Silver badge

      Re: Actually Bitcoin…

      I am 100% certain that I'm not the only regular commentard that was on the cypherpunks mailing list in its heyday. Please don't speak in such a way about matter when you clearly have first- and second- hand witnesses present.

      The goal absolutely was to create a medium of exchange that was beyond the influence (let alone control) of the existing bank-nation-state complex. There is a reason that electronic currency was followed immediately by revolution, after all. The discussions centered around small, rapid transactions. I don't even recall the idea of central exchanges being discussed, but if they had been, the idea would have been dumped on since such sites are obvious attack points for the existing Order. The model was the remailer network. The idea was that the miners _would_ be the servers that people used to promulgate transactions.

      Since for the first decade, we had no functional examples, the sheer volume of what was being spoken of was not so evident. It's been a while, but I don't recall anyone addressing the issue. Indeed, the CAP theorem was only published in 1998, and as I admitted recently, I never considered its relationship to coin until it was mentioned here a month ago.

      Of course, I cannot speak for Satoshi, but the cypherpunks _were_ the ones playing with it when it came out.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like