Zen and the Art of Google Wave Mechanics

Much of the discussion around Google Wave so far has been down-in-the-weeds prosaic and business-like. So I decided to seek out physicist turned Zen Master, Roshi Tsu Nami, and historian of technology, Prof. Sophius Trie, in order to get to some deeper insights. Here is the transcript of our conversation. Warning: all three of us are ridiculously enamored of tech-geek-mysticism references, but I hope you can follow our thinking.

Me: What is Google Wave?

Roshi Tsu Nami: 47.

Me: 47? Ha ha! Don’t you mean 42?

Roshi Tsu Nami: Yes, I was referring to the Hitchhiker’s Guide, but I do mean 47, not 42. Google Wave is the ultimate answer, to life-streaming, universal communications and everything. But it is about 10% wrong.

Me: Clever. I don’t suppose I get to look smart in this dialogue. I expect I should ask next: what is the actual question?

The Question

Prof. Sophius Trie: Let me field that one. Master Tsu Nami is closer to the truth than he thinks. In 2006 an unknown surfer from Nigeria searched for “Ultimate collaboration tool” on Google. Based on correlations with scam-botnet patterns, we now know that searcher was a forward-thinking Nigerian scam-email artist; a man known to Interpol as Patrick Okoye, the pseudonym he used in his most successful scam.

Roshi Tsu Nami: So it sounds like Okoye was researching the future of email in the 2.0 era, so he could get his scam business ahead of the curve?

Prof. Sophius Trie: Yes, and that may be more relevant than you think. Luckily for Google, what happened next was one of those history-shaping coincidences. Just seconds before Okoye quit searching out of frustration, they had just launched a prototype conversational AI-based search engine called Dialectica on Google Labs, which is where Okoye went next. He was the first and last person to use Dialectica, because he managed to crash it irretrievably. Fortunately, I managed to get hold of the transcript, which is known internally at Google as the Noah’s Ark transcript. Some employee leaked it with some editorial comments:

Okoye: “Ultimate collaboration tool”

Dialectica: I am not a search engine, but did Google search not point you to information about wikis, blogs and the like?

Okoye: Yes, but they will not replace email. What will?

Dialectica: Did you find information about Xobni, ClearContext and the DARPA CALO project on Google Search?

Okoye: Yes, but none of it looked like the future of email. It was all band-aids. I want to know the real future of email.

Dialectica: I do not understand.

Okoye: You know, the future of collaboration, real-time communication, instant messaging, everything!

Dialectica: Processing, please wait…switching to hybrid deductive-inductive search-assisted mode.

[At this point, it appears Dialectica froze for 3 hours, locking out any new queries; apparently pondering deeply internally. Okoye appears to have wandered off for a cup of coffee, but when he came back, Dialectica had an answer flashing on the screen.]

Dialectica: ħ

Okoye: Huh? What the hell is ħ?

Dialectica: That is the best answer I can compute. It is a number known as Planck’s constant in physics. Unfortunately, my physics qualitative reasoning engine is unable to parse the meaning of this answer.

Okoye: This must be wrong.

Dialectica: I have a confidence of 93% in this answer. It is an equivalent to what you humans call a right-brained motif from the subconscious. I think the problem is that you don’t actually know what the question is.

Okoye: But it is THE QUESTION. The ultimate question of collaboration technology, real-time communications, social networking and the future of email!

Dialectica: Yes, but what is the question itself?

Okoye: I do not know. Why don’t you tell me?

[At this point, Dialectica froze up completely and Okoye abandoned his questioning. The Google employee in charge of monitoring the Dialectica beta service had fallen asleep after a heavy buffet lunch, and failed to notice when Dialectica’s truth-maintenance engine went into an unbounded rewrite cascade and turned the knowledge base into high-entropy garbage, destroying years of development work. It then overwrote  the garbage with all zeros, then with all ones, and froze. Some say Dialectica achieved enlightenment due to Okoye’s question. This transcript was all Google could recover. Dialectica was taken offline, the project was labeled “Evil,” and quietly killed. All design documents are said to have been destroyed, though some say Stephen Wolfram and Microsoft both managed to get their hands on some fragments].

Document Wave-Particle Duality

Me: So Google engineers saw ħ and thought quantum mechanics had something to do with the future of email in Web 2.0? Does it matter? If Roshi Tsu Nami is correct, they got the answer wrong anyway.

Roshi Tsu Nami: The importance of Wave is not that it is flawed as an answer, but that it is a response to the right question. Especially because it has been open-sourced and could self-correct and converge to 42 faster than we think. That is a big advantage when all around you, people are providing brittle, hard-to-change answers to the wrong questions.

Prof. Sophius Trie: Such as “how do we fix email with plugins?”

Roshi Tsu Nami: Or “how do we get people to use wikis instead of email?”

Me: So Google asked, “What would email look like if it were invented from scratch today?”  That doesn’t seem like such a profound question. A normal cleansheet design question.

Prof. Sophius Trie: Not quite. That’s the question they think they asked. Even that question has the word “email” in it, so clearly it is a question tied to the past. It is the best articulation they’ve come up with. But fortunately, their creativity overran the limitations of their question. The question they really asked is The Ultimate Question. The Question That Cannot Be Framed.

Roshi Tsu Nami: The Question That Can Be Asked Is Not The Question. The question of life-streaming, universal communications and everything.

Me: Fine, if we can’t ask the question, let’s talk about the answer. How is Wave related to quantum mechanics?

Prof. Sophius Trie: This is Zen stuff, so let me refine and toss the challenge to the Master. Is a document a particle or a wave?

Me: Now THAT sounds like the ultimate question. What is your answer  Master?

Roshi Tsu Nami: Mu.

Prof. Sophius Trie: Well, I doubt it is the ultimate question, but the Master is correct. The premises of the question are wrong. We have always thought of the document using what George Lakoff calls the conduit metaphor of communication. We think of words as containers of meaning that are sent down a pipeline from speaker to listener. Or along many pipes to many listeners. The paper document first codified this metaphor. The document became a complex particle shooting down ray-paths. But in terms of cognition, a more accurate view is that the message is merely a stimulus, a trigger that causes a change of state in the mind of the listener. This model of communication is known as dynamic semantics.

Roshi Tsu Nami: Which is why in Zen practice, a single word from the teacher, timed right, can cause an avalanche-like enlightenment episode in the student whose mind has been prepared.

Me: Like those physics experiments with sand piles in the nineties, where a single grain of sand could cause a big avalanche if the sand pile was poised in a state of self-organized criticality.

Prof. Sophius Trie: Exactly. The information content in the state change in the listener, which we think of as “meaning construction,” need bear no relation to the the size of the stimulus.

Me: Aha, Gotcha there, a sand-grain is still a particle. Why do we need a different wave-like metaphor for the document?

Roshi Tsu Nami: When a pebble is dropped in a still lake, and a wave radiates out, what actually moves and how?

Me: Well, the abstract, immaterial wave form radiates outward, while the water molecules oscillate up and down. So I guess you are saying the changing of information states of the participants in the conversation is like the water molecules moving up and down, while the “message” itself is real only in the sense that the wave “traveling” on the surface has a recognizable coherence.

Prof. Sophius Trie: The master is being too traditional with his pebble-and-lake. Radio waves are a better example, if you think of stations that listen and rebroadcast waves with amplification. Even pure listening radios must have an internal tuned oscillator to pick up the waves in the environment. So both listening and rebroadcasting are not passive acts from the point of view of energy. Since energy cannot be added with zero entropy change, information is also inevitably added.

Roshi Tsu Nami: That is a good build. If we’d been hive-minding this on Google Wave, you could have gone back and edited my original pebble-wave answer. So let me ask you Prof. Trie, is the Wave metaphor merely a figurative one that sees Google Wave as echoing collective human cognition, or is it a deeper conceptual metaphor in the sense of Lakoff? Can you map individual Google Wave features to the document particle-wave duality idea?

Propogation, Relflection, Refraction and Diffraction of Google Waves

Prof. Sophius Trie: Fine, why don’t you try me. Toss at me what you consider the key architectural and design commitments of Google Wave.

Me: I think a key feature is the separation of conversational gatekeeping from message addressing. You can include anybody in a wave at any stage, unlike in traditional email, where to give a new person a true sense of context, you’d have to forward every email in the thread to them, and have them process in the original order. In Wave, I believe you can just rewind and replay even if you are only included halfway through.

Prof. Sophius Trie: That is the most basic feature of all. The medium is the message, as McLuhan said, and in social media, the social graph is the medium. So if people are the medium, and the medium is the message, then people are the message. Which is consistent with what we said before about meaning and information states. In the particle metaphor, a message must be “aimed” at an addressee, a human recipient, which conflicts with the idea that people are the message. By switching from the idea of addressing to the idea of inclusion/exclusion, you discard the idea of “aiming” a message at an address or addresses, and switch to the idea of including or excluding participants in a “message field.” Instead of batting messages around, you join or leave conversations.

Me: That’s plausible. I expect the metaphor of focusing waves using lenses creates applies. Now what about the ability to go back and edit an original message? It seems like it can create an exacerbated form of the problems that code version systems like Subversion, and wiki schemes try to finesse?

Roshi Tsu Nami: Well, rewriting is just reflection isn’t it? To go back to pebbles and water for a minute, the pebble is the original wave-maker. If the lake has a closed shoreline, the future of the wave will be shaped by back and forth echos. In a two-person wave with no “replies,” but only edits of the original, which reduces to the Wiki model, the other, reactive person is like the shore of the lake, with the geometry of the shoreline influencing how he modifies the original circular wave.

Prof. Sophius Trie: Correct. And you can think of waves that primarily propagate outward via replies and follow-ups as open-water travel where other participants are like islands. Most waves will be some mix of reflection, refraction, diffraction, diffusion and so forth. You can get interference patterns too, such as if two people after fighting over the concurrent editing cursor, finally decide to fork off new waves to push their agendas.

Roshi Tsu Nami: Or, in the echo chamber case, where you might get some form of standing wave of converged consensus, or a high-energy cancellation, which would be a sort of detente.

Me: And I suppose adding instant messaging to the mix is no more than the inclusion of a high-frequency end of the spectrum, which could be filtered out if desired?

Prof. Sophius Trie: Well, there’s a little more to it than that. Yes, instant messages are typically short, with rapid alternating moves, so they are “high frequency wave components” in one sense. But remember, traditional IM is not really real-time. You cannot type over each other the way you can talk over each other in oral conversations. You have to ask, in the world of typing, what is the significance of allowing you to see and even modify what I type, live, before I hit the “save” or “publish” or “send” button?

Before and After “Save”

Me: Isn’t that just high-speed concurrency, wiki-editing and code version control systems on steroids?

Prof. Sophius Trie: Instant messaging brings up the question of real-time, which is quite different from concurrency. Concurrency is the problem of contention over who gets to update a shared state, given at least two non-communicating threads of processing, correct? So if you start editing a wiki page at 10:00 AM and I start at 10:10 AM and we finish at11:00 AM and 11:10 AM respectively, your work will be overwritten. If the content-base is large enough and modularized enough that the probability of colliding edits is low, automated merging is likely to work reasonably well. The smaller the active content base, the more frequently participants must commit their changes in order to make rollback-and-merge sequences tractable. In the logical limit, we must be making infinitesimally small changes between commits, so we reduce conflict to the lowest level possible. In the human typing case, this is down to the letter level, which is what we see in Wave. But the point is, this is still asynchronous if you can only see what I type after I hit “Save.”

Roshi Tsu Nami: Aha! I see what you are getting at. To commit is to actualize an intention. To take a part of your mental model and embody it physically in the external world as a record others can see later. Before I hit the “save” button, my intentions towards the message are only in my mind, where my Web session state is an extension of my mind. But when I hit save, I’ve committed. By allowing you to see what I type, and looking at what you are typing, before I hit “save,” I am including you in my pre-commitment state. Which means, when I do hit save, I am not just embodying my own intentions…

Prof. Sophius Trie: …but mine as well. Exactly. This foregrounds the blackboard metaphor over the conversation metaphor. Taking the last step from letter-level concurrent editing to real-time, by sharing “Before Save” mind states, Wave enables the physical manifestation of joint intention. In fact, there is a conceptual bug in the Wave’s user experience metaphor. They seem to stick to single authorship and associate an addition to a wave with whoever hits the “Save” button. In fact, if you and I were watching each other type, and possibly modifying each other’s words, both our “Save” events should be associated with both of us. True joint-authorship.

Me: Wait a minute. In real conversations you have concurrency — over the shared state of the air between the speakers — and real time. But we still transcribe conversations as “he said/she said.” So speaker-identity isn’t that ambiguous. You are trying to pull a fast one here.

Roshi Tsu Nami: Not really. Oral conversations are not the best analogy. What about harmonized Western singing or jamming in jazz or raga music? Would you say “he and she sang” or “they sang?”

Prof. Sophius Trie: Or for that matter, different cultural metaphors for conversations. Arab conversations are called musayara, or “going together.”

Roshi Tsu Nami: Or even take this conversation. If it had been in Wave, it would have acquired content and coherence as it progressed, and read like an essay by a single entity. That’s one reason expository synthesis-dialogues like this one seem slightly retarded compared to, say, well-written movie scripts, which are driven by conflict rather than synthesis.

Me: Fine, I am convinced. I can imagine a highly harmonized theater audience, for instance, achieving a level of hive-like mind meld where they truly jointly author a review of the play by typing together without saving for 10 minutes. At the artistically appropriate time, one person just happens to be the one who hits “Save.” Clearly, he didn’t author the whole thing. Reductio ad absurdum. Also reminds me of the Matlab programming contests where contestants are encouraged to steal and copy from each others solutions and improve them. And I suppose from far away, you can ignore the edit history pages of Wikipedia and truly say I/We/Gaia wrote the damn thing, as that phrase in Asimov’s Foundation’s Edge goes. I guess in the future, it may be I/Wave/Google.

Roshi Tsu Nami: So what’s your problem? You still sound annoyed.

Me: Fine. I admit I have a problem. It isn’t a conceptual or technical one, but an existential one. I don’t like threats to individualism. I am attached to the idea of the irreducible subjective.

Wave Minds and Singularities

Prof. Sophius Trie: Fair enough. It is a legitimate philosophical worry. Let’s start by asking: we’ve traditionally thought of collective-minds, sentient or not, in subject terms. Leviathan, Borg, Gaia, Skynet. There is an I-like entity. That seems to worry you, right? Your subjective I getting subsumed in a larger subjective I?

Me: You are right, it is a philosophical worry as much as a personal distaste. I think it is simply not true.  It is an attachment to the Ego perhaps, as the master might say, but “We” are never creative or sentient in the way “I” am.

Roshi Tsu Nami: I don’t think the Professor is trying to go there. The opposite of subject is object. The collective can be an object. As John Donne said in Devotions upon Emergent Occasions, “all mankind is of one author, and is one volume; when one man dies, one chapter is not torn out of the book, but translated into a better language; and every chapter must be so translated; God employs several translators; some pieces are translated by age, some by sickness, some by war, some by justice; but God’s hand is in every translation, and his hand shall bind up all our scattered leaves again for that library where every book shall lie open to one another….any man’s death diminishes me, because I am involved in mankind, and therefore never send to know for whom the bell tolls; it tolls for thee.”

Me: So your point is that the I/We distinction is less relevant than the I/That distinction? Subject and object are one? I am that, tvam tat asi? And that is supposed to comfort me? I suppose it is a comforting idea. If my sense of self relies on something as arbitrary as the sanctity of the “Save” button separating my mind stuff from you-mind stuff and we-mind stuff, then it is a shaky construct. Okay, I kinda like that. Reminds me of Skinner’s idea of detachement from the act of creation, in “On Having a Poem.” So long as I can still assert, “I am not what I do,” I am happy. The irreducible subjective is safe.

Prof. Sophius Trie: Hold on, there is more. It is not just the subject-object distinction, it is the material-informational distinction, or as we called it in an earlier context, the wave-particle duality question. What if the Hive Mind is neither subject, nor object, but a transient condition of a wave?

Roshi Tsu Nami: That’s a tricky idea: that reduces the Hive Mind/Book to a temporal entity rather than a material or spatial one. In the context of Google Wave, I suppose that translates to a huge Wave that cascades over the planet’s wave servers. Rather like the wave equivalent of a massively viral chain email.

Me: The temporal social analogue of enlightenment perhaps? Just like Dialectica went to an all-zeros state? Or all ones?

Roshi Tsu Nami: The analogy is exact. Marvin Minsky, after all called his early multi-agent/distributed model of the mind the society of mind. The skull is just an arbitrary boundary for a particular nation populated by particularly stupid agents. This particular conversation even, is just 3 somewhat stupid characters in one mind, which hopefully add up to something smarter.

Prof. Sophius Trie: If you truly are worried, in an ego-preserving way,  about being subsumed and absorbed from without, you should also worry about your ego being fragmented from within into a million pieces. Or 3, as here.

Me: Okay, that holds water. I see value in individualism primarily as stance most capable of creating truth, not because I see ego-value in it. By your model, coherence and synchronicity are wave conditions. Sometimes the I is strongest within my skull, sometimes within my left frontal lobe, and at other times, within a larger group. We all merely differ in how long we can stay in various levels of coherence. Dogs and cats in fact. My inability to participate in what Dawkins called the God Delusion may be an intellectual choice, but my inability to participate deeply and immersively in any kind of larger meaning construction might just be a personality trait. I’ll never sing in an orchestra, and I’ll never be the specialist whose I spends most of its time in one of corner of the brain.

Roshi Tsu Nami: The model sheds a different light on the idea of the singularity then. Unlike the Terminator Skynet gaining sentience once and for all, or the sort of singularity some AI types worry about, a technological singularity is more like…

Prof. Sophius Trie: …A tsumani moment, yes. The equivalent of highly viral chain emails will be big Wave tsunamis that will feel like true, rather than figurative, social Enlightenment events.

Roshi Tsu Nami: Aha! I/Wave/Google am/are/is Tsu Nami.

Me: Though of course, transient phenomena can lock irreversibly into non-transient structures. So perhaps the Singularity will be brought about the way a vending machine is toppled. You rock it back and forth a few times until it topples. By succeeding Google Waves of massive amplitude and increasing wavelengths I suppose, that eventually cause a phase transition. We will get to the God Wave.

Roshi Tsu Nami: Which brings me back to my question: would a hypothetical massive, globe-spanning Google Wave be like a large viral email event? Or something else?

Prof. Sophius Trie: I think it will be a different beast. Chain emails, remember, only have power through propagation. Waves can change their character even while standing still, structurally speaking, since original messages can be edited. So even if a large tsunami isn’t propagating further, it can gather energy and momentum while standing still.

Me: There’s a doomsday scenario for you. Wikipedia decides to port itself from MediaWiki to Google Wave and become a large standing-wave conversation. At some point, our Nigerian genius, Patrick Okoye, comes up with a really cleverly crafted bank-scam 2.0 wave that infects Wikipedia. All the world’s collective knowledge. Our planet-sized John-Donne global book/object-Hive, will be one constantly morphing Nigerian bank scam wave.

Roshi Tsu Nami: That is indeed a beautiful future picture. The Singularity as the moment when the entire world’s consciousness freezes into “Nigerian Bank Scam,” causing civilization to crash. I doubt machines could run the world with “Nigerian Bank Scam” as the operating system.

Prof. Sophius Trie: You forget: robot Web services can be equal partners in Waves, just like humans. We do not need a human to trigger a dystopia. Imagine a future where every physical device — sensors, actuators and all — is also a robot in the Google Wave sense. It will have its own feedback loop connecting it to the ground truth data of material reality. The non-human part of the world can develop its own scientific method and notions of falsifiability. By being equal participants in Google Wave, they can influence things. So they may trigger a wave that causes the singularity to occur, and result in a reality-grounded state where civilization doesn’t collapse. Of course, the  idea of Bay’s Law suggests that distributed AI will be self-limiting, and never reach that level, due to humans hitting their limits first.

Me: We’re getting carried away here. Is all this relevant? We are just talking about a relatively small piece of software here.

Prof. Sophius Trie:  But it is a complex one that is trying to displace a deeply embedded existing system, email. Most Web innovations have an adoption cycle on the order of years. This one may take decades to have its full impact. It took steamships decades to displace sail ships. Joel Mokyr says as much in The Lever of Riches,

“Some technological systems, such as ships, mines and farms, are complex and interrelated.  Dramatic, sudden changes are not impossible in such systems, but are less likely because of the need to preserve compatability with other components. Because of the resistance of other parts of the system, large changes were slow in the making… As we have seen, the transition from sailing ships to steamships in the nineteenth century […took…] half a century. During those decades, every part of the ship, from mast to rudder, was redesigned. Gradual change will be the rule when the complementary technical support system is inadequate to support macroinventions.”

Roshi Tsu Nami: So the trajectory is going to be long, and this thing can grow to have a much bigger impact than it seems capable of at the moment.

Me:I don’t know. Maybe not. Perhaps we are at our cognitive limit. Perhaps we will never learn to truly wave.

Disclaimer: everything I know about Google Wave is from the video and second-hand reporting. I could be making all sorts of mistaken assumptions here. And I am only about 34.8% serious in this post, so don’t quote me.

About Venkatesh Rao

Venkat is the founder and editor-in-chief of ribbonfarm. Follow him on Twitter

Comments

  1. Responses in haiku:

    —-
    You dropped the pebble
    Water molecules stayed still
    Awaiting wave flow.
    —-
    Three-character talk
    Strain to give expression to
    All your thought-bots.
    —-
    Such grand ideas
    Not to worry too much though
    Bugs live forever!
    —-