Google Multivac | Hacker News Reader

Google Multivac

neil.fraser.name

57 pointsvijaydev17 comments

I think one thing that always fascinated me was how anchored Asimov was in reality. I began reading 'The Gods Themselves' which Asimov wrote as a response to another writer (whose name escapes me) using a Isotope which couldn't exist and ignoring the science related ties that it would mean.

His science fiction was always speculative, but in as many places as possible it was anchored to the reality he knew.

Simple amazing.

Oh, and yeah, Google is scary how big, powerful, and... omnipresent it has become.

Cool, maybe Google will figure out how to reverse entropy.

Actually MultiVac sounds closer to Wolfram|Alpha, especially the data being curated by civil servants, and only subsequently analyzed/interpreted/correlated by the computer.

That's really cool that we're implementing a lot of the sci-fi. What hasn't been done yet? What bits of Multivac don't exist at the GOOG yet?

The ability to answer The Last Question: http://www.multivax.com/last_question.html

Well, the teletype interface used throughout the robot stories, for one. But you could probably rig that up in an afternoon with a daisywheel printer, acoustic coupler, and just a little bit of tty hackery.

The bit where it understands what you want and gives you one single helpful answer instead of dozens of pages of links to forum posts, each of which contain a clone of the same unanswered question as all the others, followed by some Google adverts.

That's much more profitable than a single answer. Jeez :D

The central plot of "All the Troubles of the World" is that Multivac was predicting who was going to murder whom. In terms of crime prediction the closest Google has come is mashing up maps and crime stats to show which areas of town are problematic. Unlike Multivac, Google has no interest in being a police officer. Whereas Multivac was built by the government, Google was built by a corporation.

> http://neil.fraser.name/news/2009/multivac/patrol.jpg

I have to wonder - is this guy wearing headphones because:

a) Google's a hip and groovy place to work and he is allowed to listen to whatever music he wants as long as he gets his job done.

b) It's noisy in there and he needs ear defenders.

c) http://www.marshallbrain.com/manna1.htm begins at Google, and that's a wireless headset connected to a Google automated manager/instruction system. Please take one server unit from repair area D2. Thank you. Go to Server container N14. Thank you. Please identify yourself and enter. Thank you. Please locate rack 8, on your right. Thank you. Please identify the red and black power lead, and unplug it. Thank you. Please locate the network lead and unplug it. Thank you. Please slide the server out and replace it with the repaired one. Thank you. Please reconnect the network and power leads. Thank you. Please take the removed server back to repair area R2. Thank you. Please report to task room 3 for your next task. Thank you.

Most large and dense server rooms are quite noisy. Ear protection can make them much more bearable to work in for long periods of time.

Yeah, there's no wires going to them. They're ear protection.

Edit: Nor antenna. Probably not wireless headphones.

Most Bluetooth headsets have their antennas inside, so that they aren't visible.

Ear protection is the obvious answer, and I included it in the options. The other option is the one you might want to believe (free choice of music). The last one is suggested as a what-if alternative. Partly because it's an interesting story (at least the first part about the Mana system is), and partly because if anywhere is going to implement a Mana-like system, Google has the drive to do new weird controlling things, the technical expertise to have a massive central task and employee management system, and a their simplified yet very large server room setup sounds like a good place for that kind of thing to be useful to cut down on skilled worker requirements.

Saying "probably ear protection" is rather missing the point of my post, and getting three upvotes for it suggests that I wasn't at all clear enough.

Point being: imagine the kerfuffle if it turned out he was being guided by an automaton task master using wireless headphones and systemic feedback from many smart sensors in the style of the coming Google power grid monitoring system - picking up a server from repair bay D2 detected, entering the server room detected, sliding a server detected and the server's details taken by RFID'd and putting a new one in, details from RFID, booted and configured and brought online automatically, old server returned and asset tracked.

Google already design their own servers and server rooms, they track power usage and memory errors and disk failures and have kernel modifications to track resource usage by program. It's not that far out, and is the sort of major thing they might do quietly which might only leak out initially by a slightly odd picture or a disgruntled insider comment.

"c" was reality at a place I used to work. They had an old IBM mainframe and a huge tape archive going back to the 60s. There were several "silos" with tapes that were frequently needed arranged inside, where they could quickly be loaded into readers by robot arms. Less frequently needed tapes were organized outside the silos on shelves. The mainframe managed the tapes itself, but did need the help of a few humans whose entire job was to read printouts requesting tapes to be swapped in and out between the silos and the shelves and go fetch them for the mainframe.

Reply on news.ycombinator.com