20110501-Hewitt, Meijer and Szyperski: The Actor Model (everything you wanted to know, but were afraid to ask) | Going Deep | Channel 9

Robust11.org  symposium on inconsistency robustness 2011



actors let us deal with consistency via local arbitration, but we cannot model global consensus. 

so how does consciousness work?  It's a representation of global consensus.  which is pretty much what attention is.  a representation of things related to self. it's local manifestation or representation or "assumption" of global consensus.

Carl Hewitt  http://carlhewitt.info/



with many cores, the programmer does not know where the code actually executes.  even what kinds of processor runs it.   so there needs to be a higher level abstraction (that goes from high level to low level).  



in a many core model, goto is fine, it's assignment that is the problem (which comports with representation theory.  to keep the assignment in place, it needs to be "held" across a chain of representation.   goto is instead, the chain to the assignment.  easier.  because we can just initiate a cascading message to stable actors to elicit the assignment.   


A COMPUTABLE UNIVERSE 

Understanding Computation & Exploring Nature As Computation Hector Zenil (editor)

(member of the Turing Centenary Advisory Committee)


an Arbiter


petri nets fail as model of physics (well, not really, not for chemistry) because 2 inputs become one output.  (20140311: but this is what happens representationally.  the failure in chemistry is that we treat small changes to a molecule such as a protein acquiring some small molecule and changing shape as a kind of state change.  that the resulting molecule is still the same molecule with some small change.  This is a category error.  The molecules are different.  it is just as correct to say the small molecule is changed but still the same when the big molecule bonds to it.  this is representational bias.  there are three molecules. a big one, a protein.  a small one, like glucose, and when they combine there is a third molecule.   in actor terms.  the two actors combine into a third actor.  the actors take each other as messages.  this is the basic conundrum resolved with the basic model and computational chemistry)


messages get passed in-deterministically.  if you want to synchronize messages, you need another actor



An actor gets a start message and it sends a go and stop message.  the go message comes back and it increments a counter and sends a go message.  this recurs until stop is reached, at which time it sends a counter message.   the stop can take in-determinately long.  the counter may be at 1 or 99 when the stop is received.  THERE IS NO REASON THE STOP MESSAGE MUST EXIST EVEN ON THE SAME ARCHITECTURE!  the stop message can be a stigmergic message sent out to an external environment and is not received back until it is appropriately returned.   note that in this model, the messages are sent to other actors addresses.   or in a biochem setting, the message is sent down the axon, far away from the dendrites.  the neuron must receive a stop, the actual atoms may be different, it is the same message.    we can do the same thing with mark making in the environment, that 'produce' stop signals in our own brains.  it's message passing, and levels of message passing.  





Address != identity.  

Actor model:  
0) It Processes
1) it stores
2) it communicates

actors exist in systems. 

Axioms:
0)  An actor can create more actors
1)  An actor can send messages (to addresses)
2)  An actor can designate next message action

I would add another axiom. 
3)  An actor can seek an address.   It does this by sending messages to actors to get addresses they know, and then sending messages to those new addresses until it finds an address it wants.   it's how we look things up on the web too. seeking/making connections

conceptually one message at a time.  But there is no reason to treat all messages in toto (an automata model).  so that all messages are treated a one message by the actor. 
messages are sent and received randomly.  the address space of an actor  or   actors receive random messages. 

note, many different actors could have the same "address" (receptors) and receive messages non-deterministically (arbitrary)  and then address is not identity.


Hewitt, Meijer and Szyperski: The Actor Model (everything you wanted to know, but were afraid to ask)

Download

How do I download the videos?

  • To download, right click the file type you would like and pick “Save target as…” or “Save link as…”

Why should I download videos from Channel9?

  • It's an easy way to save the videos you like locally.
  • You can save the videos in order to watch them offline.
  • If all you want is to hear the audio, you can download the MP3!

Which version should I choose?

  • If you want to view the video on your PC, Xbox or Media Center, download the High Quality WMV file (this is the highest quality version we have available).
  • If you'd like a lower bitrate version, to reduce the download time or cost, then choose the Medium Quality WMV file.
  • If you have a Zune, WP7, iPhone, iPad, or iPod device, choose the low or medium MP4 file.
  • If you just want to hear the audio of the video, choose the MP3 file.

Right click “Save as…”

At Lang.NEXT 2012, several conversations happened in the "social room", which was right next to the room where sessions took place. Our dear friend, Erik Meijer, led many interesting conversations, some of which we are fortunate enough to have caught on camera for C9. We'll begin with these Expert to Expert episodes with a "standing" conversation (participants stand comfortably close to the whiteboard) with computer scientists Carl Hewitt, Visiting Professor at Stanford University, creator of the Planner programming language, inventor of the Actor Model (the topic of this conversation),Clemens Szyperski, an MSR scientist working in the Connected Systems Group and Erik.

What are actors, exactly? No, really. What are they? When is an actor an actor? Everything you wanted to know about actors, but we're afraid to ask... It's all right here. Big thanks to Carl, Clemens and Erik. This is an excellent E2E(2E)! 

Tags:

 

Follow the Discussion

previous