Sorry that I missed it!

We are an open, voluntary collaboration.  We do not generally seek funding or offer any vestment vehicle, as we find it antithetical to the aims of the project.  (You are of course welcome and encouraged to support the project in other ways, including funding individual developers directly at your discretion - we just don't promise any sort of monetary ROI from the project on such.)

---

Agoras is Ohad's baby, but Tau was always a separate, distinct, and ultimately unrelated project.  All of the other developers were always against the very idea of Agoras being started, and basically begged Ohad not to do it under the assumption that it could only cause problems for the project.  (It did.)

You can find a lot of detail about our effort on our wiki (linked two posts above yours) and can find even more by coming to our channel and asking questions.  Personally, I think our directions are about as interesting as things can get.

---

We would be very interested if anyone can show a sound reason why the concept should not work.  We've had an open call for such for years.  So far, noone has actually demonstrated any flaw in the premise.


There was really a break-up.  Ohad really did leave the project to do "something else" (we still don't know exactly what) which he calls "new tau." (Hence our nickname for AN being "oldtau" now...)


See wiki link a few posts above...  

Ohad's original whitepaper was about the original design, and so should still apply to our work, as we have not changed plans as he has. (As far as I am aware, he has not published any updated whitepaper/explanation for his new logic, so I can't meaningfully speak to what those changes are.)  We keep a sort of roadmap in our wiki, which is what mnemonicsoup linked to specifically, but we do not make guesses about when we will launch.  (There are some other resources which can be found in the links in our IRC channel's topic and something like a "working project" on stoop's github - though it is likely not what you're expecting....)

If you are interested then please come talk to us, we'd love to explain what we're trying to do and what we're all about to *anyone* who cares to get involved!

The Autonomous Nomic Overlay Network
aka Autonomic
aka A.N.O.N.
aka oldtau

We are all of the tau developers, less Ohad, continuing with the original project plan.

You can find us in #autonomic on freenode IRC.

I've been asked to address this post directly.

First, a small disclaimer: I post this as a personal response, not as a representative of the Autonomic project.  This post has not been discussed with the others, and does not necessarily reflect the views or official positions of our project.


Of course there are such expressions.  Ohad's refusal to admit this (even contradicting himself by giving both answers) remains baffling to me.

Here is what is commonly considered one of the "canonical" examples, and it is commonly used to prove various meta-theoretical differences of second order logics:

This is the class definition of finite structures.  It has no mso or fol equivalents.  Its' negation is the class definition of infinite structures, and ofc also has no mso/fol equivalents.

We've covered at least one or two dozen such examples in the autonomic channel over the past few months.  Lately we've been discussing the expressibility boundary between full second order and Henkin semantics, as this is particularly relevant to the decidability of logics above first order.  In that context, here is my favorite working example, due to its' amazing simplicity in illustrating:

This is logically valid in full second order semantics, but does not collapse to first order as valid.  Under Henkin sol semantics it is conditionally satisfiable.


Note that his given example does not meet the request.  It is a first order program, with unlimited/unconstrained inputs.  His verification does not assert any property of the IO relation, for any inputs - it only verifies that the function is constant.  (I agree that this property is first order expressible...)  

He also seems to conflate order of functions of a signature with order of logics over a signature.  (He doesn't seem to understand order of functions, either, for that matter...)

(EDIT: It was later pointed out to me that Ohad made a second post with a second example expression which was a higher order function.  It still has unlimited/unconstrained input, asserts no property of IO relation, and he still verifies only a first order property of the output (as he points out at the end of his post) so this is also not an example that meets klosure's request.)


I'm afraid that I can't speak much to this one.  Looks like nonsense to me.


You can find significant discussions of these topics in the #autonomic logs.  I'm not following ##idni anymore but am told by others that it has devolved into a weird sort of echo chamber, mostly about (very old) solutions to transform problems and with very little to do with any particular logics.  I'm guessing you would not find any clarifications there.

I'm also told that Ohad has now given up on mso entirely and moved to more expressive logics.  From what I hear it sounds like he is lately exploring the region around/above the P-space logics, and in the "murky" area between the exptime algorithms and primitive recursion.  (It is still unclear to those I've spoken to what the logic he "has in his head" actually is.  AFAIK he still has not done anything to make concrete (or even further describe/qualify) his proposed framework.)

I like to joke that he is slowly making his way up the complexity families, and may even soon rejoin us in exploring "true" omega-order HOLs with full semantics, covering the full arithmetical hierarchy and in the existential N*N fragment of the analytic hierarchy.
(I can only hope that, along the way, he remembers/rediscovers why completeness and deductive proof inference should be considered problems for the system, not features.)

As far as I'm concerned, the debate about "mso v mltt" has become entirely moot.  There still seems to be room for debate wrt: completeness, decidability/term-inference, Lowenheim-Skolem/compactness/Lindstrom, (general) consistency of collapse of expressions between orders, consistency of LEM as axiomatic, or the necessity of general expression within the system.  He still must disagree on one or more of these points, otherwise he'd inevitably have to just be back at MLTT as the only possible option - and back on track with the "oldtau" plan.

If he did "come around" it is unclear to me if our A.N.O.N. project members would welcome him back with open arms or tell him to go jump.

Personally, my position remains that all are welcome.  Unlike newtau lately, we do not generally seek to moderate ideas or hide our design-detail information.

Edit: fixed typos

Sadly, yes.

Even I have to agree that this is some seriously questionable report, here.... HEH. 

(Ok, I'll bite...)

I'm very curious: what "command" do you believe I've given... and to whom, exactly?

Right, this is one of the reasons for the small/subtle changes in the payouts model.  Under the model I propose, the attacker would need to re-submit his complex job after each solution he issued to himself.  (His job would also have to be "actually" complex, and not just be full of "meaningless" busy-loops which do not contribute to output value, mind you.)  This creates an opportunity at each solution found for miners to reject his new jobs (or up tx fee for them) if they feel the network becomes too loaded.  At that point the problem becomes no different from an attacker trying to stuff up the btc network with dust transactions filling up the block limit - he can still try to do it but miners can just make it ever more costly for him until he is broke and must just stop because he can no longer afford it.

The best ways to keep up on the project, at this time, would probably be to follow the irc log and maybe keep an eye on the wiki.  Join #AutoNomic on freenode and check out the links in the topic for more info.

Also, just come talk to us!

Seriously, we answer questions... for free... all day.  We even delight in doing so.  All anyone needs to do is come to us and ask them.  We have nothing to keep secret, nothing to sell, no strings to attach, and we want as many people as is possible to understand (and perhaps even participate in) our direction.

This may seem like unusual behavior in this contemporary space of crypto projects, but it is how we've always felt that it was best to choose to operate.

Even shorter version:  Instead of giving a machine and getting an input/output pair, you give a model of a machine's behavior, and get back an input/output pair's relationship to that machine.  Instead of having a chance to find a PoW certificate at the end of each work attempt (effort of which varies run to run) there is chance to find a PoW certificate for each individual value propagation in the program, which would be uniform both run-to-run and job-to-job.  A job requiring on average twice as many assignments per attempt would generate twice as many work proofs on average for the same number of attempts run.

The very short version:  A job requester would annotate an SSA form program (in a specific machine model resulting in a particularly structured (binary) flow graph)  with a simple liveness/reachability model so that miners could (quickly, and without running any example case inputs) verify the necessary complexity bound of the job's individual work task before selecting.  PoW solutions would operate a little differently (still using "per user" generated inputs incl nonce data, but basically hashing/checking "instruction by instruction" instead of at the end of each input run) so that PoW solution rates become uniform across jobs, being able to be found at any point mid-execution.  (PoW prize pool would probably also need to work a little differently, with any amount of proof-of-work certificates able to be submitted before a bounty is found, and the PoW pool being divided proportionally after.)  Bounty solutions would include an annotation of the original model with information about the eventual I/O relation, such that verifying the output submission can be reduced to an instance of a satisfiability problem.  Nodes (all of them) would validate solutions against this model. (They would still need to "re-run the program" by a symbolic interpretation, but could know that they are doing so in an optimally efficient way - effectively skipping any "unrelated loops" encountered.)  Jobs would always end after one bounty is found.

I'm summarizing a lot, but that is the basic idea.

Yes, I just use the generic, age-old p2p terminology for any form of systematic recentralization, regardless of the trendy term-of-the-day for it.  (... which seems to vary month-to-month anymore.)


I doubt it is something that would get sorted out before you guys proceed with the lite wallet launch.  It isn't entirely clear to me what the XEL transition plan is from there, though.... so maybe it is something that could be done after or maybe it is something where we'd be talking about a new, distinct network.

Source is still in the form of experiments scattered between our github repos, pastebins/gists, etc.

We don't really spend time on things like website, marketing, "PR" but focus only on our work to finish the design and implement.  (We aren't selling anything so we don't market anything...)  We do have a fledgling wiki which has some overview information, and of course our public irc log.

However, the best way to learn about our project is to just come talk to us about it and ask questions.  We will do everything that we can to try to explain it!

I was never aware of any expectation of being a gentleman.  Our agreement was simple, I would explain the design and do my best to answer any questions, and Ohad would implement - prove the concept.  I describe, Ohad programs.

This was the explicit agreement we made at the start.  I'm a little baffled as to why it was a suitable agreement for years, but is now today suddenly "non-gentleman."

All I really want to know now is if you agree that there are second/third/higher order theories which can not possibly collapse to be consistently proven in monadic second order?

If you say "yes" and agree that there are such statements then we have some common ground to perhaps proceed with.
If you say "no" (as your other comments have implied) then your perspective differs from that of the past ~60 years of philosophy, and I'd surely like to understand how/why.

If you give me both answers and then refuse to say another word in clarification, what am I to make of your behavior?  Inconsistent?  "Non-Gentleman"ly?

Anyway, all I was doing was answering some questions about mltt and autonomic.  I didn't come here to revisit your nonsense yet again.

Thanks! It is nice to be noticed!  

I've been lurking, and still following the progress of XEL closely week to week.  Unfortunately I can't say that I've been particularly happy to see where it has been headed, personally.  IMO the decision to partially re-centralize with bonded master-nodes and the decision to launch a "lite wallet" token are both likely mis-steps.

Having 0 stake in XEL it is obviously not my business, but I had held high hopes... and now I have little but worry for E-K's work ever making it out into the world in a form that is anything like what he would have really intended.

I've been thinking a lot, lately, about a the idea of a relatively simple variant of the earlier XEL design which might address several of the concerns with the model, particularly with regard to the PoW rewards.  It would eschew a dedicated programming language, in favor of a machine model, but at the same time make for a more consistent work calculation and allow for some additional assurances.  It may not be something I ever pursue further but I am ever more convinced (though still not quite 100%) that the whole thing might even actually be possible after all.

You asked.


We have been going on with development for a year.  We decided it did not make sense to make a redundant announcement and white-paper considering we make no change to original plans.


Who's explaining how wrong Ohad is?  Are you referring to the quoted posts from Sept/Oct?

Apparently, no-one is still discussing.  Stoop tried just today to get a "math focused" discussion going between the three of us.  I asked Ohad for his belief wrt a particularly relevant logical fact about collapse of orders, a yes/no question.  He gave me both answers, then the silent treatment, more or less.  I'm not sure how we can hope to discuss with someone who can not even consistently tell their position.


No, but when the logic does one should at least take pause.

Autonomic is (literally) all of the developers who worked on "original tau" except Ohad.

We started a new project "Autonomous Nomic Overlay Network" which intends to carry out the "original tau" plan.

The project exists primarily in #AutoNomic on freenode.

Shen (Qi II) is CLOS under MLTT and so is much closer to the language of AutoNomic than of newtau, no?

A photo finish!

Gz loracle.

Gz also to EK, as we both really just brought his original algorithm down to limits for the problem.

Now we need a contest to make a more reasonable testbench for another contest?  This just turned into a micro-optimization challenge...

NM, here is the best I've gotten to so far.  We'll call it a study in realistic simulation being hard and overfitting being too easy.  Somewhat ironically, it is your original algorithm with different parameters and basically only one (subtle) change.

http://pastebin.com/YD6bYzyP

This beats "similar optimization approaches" for my few attempts at a hand-rolled re-target, but I could probably score better with some more time spent there.  I'm curious to see if anyone can beat it, as well.

Let me know what you think.  

Yah, this testbench is not so great. 

Using a variant of EK's alg, I reached a largest error of 2.52601249956 on the reference seed.  Without cheating, this time.  Is that good?  I think I still have room for improvement, but I am getting tired.

EK: Can I submit my entry by pm?