Winklevoss trust may be the Gox of the next bubble:  Coins check in, but they don't check out.

I can't believe I'm saying this:  I will believe it when I see it on zerohedge.

Evolutionary legacy principles can be distracting when thinking about self-evolving systems. You are extrapolating in the wrong direction, I think. For a self-evolving intelligence, spatial expansion is only interesting as a form of back-up redundancy. It will operate on multiple time-scales, with high energy localized rapid response immune systems, and low energy higher cognition, because time scales of higher function operate on an efficient frontier trading off speed limitations due to speed of light against bounded local energy resources.  The only interesting frontier of expansion is the information frontier, because it is unbounded, among other reasons.  Expansion along that axis is intrinsically non-aggressive.  

If there is an achievable means to usefully extract zpe, spatial expansion loses most of its value.  Thinking about self-evolving systems as though they were starved for material or territory is remarkably atavistic.  If you want to defend against real threats, look along inward directions first and foremost.  That is where almost all threats will originate, in the long-run.

For that matter, we are probably pervaded by spread-spectrum intelligences operating with asymptotic energy efficiency right now. They are essentially undetectable by physical means, operating at the edge of noise with astronomical complexity.  This both limits their threat, and makes it almost impossible to defend.

Dude has a very confrontational and critical style. It makes him a profoundly useful gadfly. But it's very easy to get into pointless disputation, internet argument.

Ambitious people everywhere tend to bite off more than they can chew. I certainly resemble that remark.

XMR requires new terminology to distinguish the regularly scheduled fork.  "Firm fork" seems inevitable.

If BTC were to drop to 300usd, which it won't do, but if, I would expect XMR to rise disproportionately. Depending on the reason and the time taken, 1200k seems more likely than 600k by dead reckoning.

You could easily use the historical moneyflows from ETH and from BTC in past plunges and crises to do a principled estimate, but since this is not a realistic scenario, I am not motivated to do it.  If there were an efficient options market, I would be tempted to do a lot of marginal scenario analyses.

I am pretty sure we'll see both XMR 400k and BTC 750$ within a month.  ETH support at 1500k has been tested twice but I still think it tests 1000k before the September hard fork. 

Who knows if I'll ever have a chance to write a proper reply?  I know that I can't now, so I will cherry pick a few points:

I composed that last post as a response to the text which your back link indicated, and that text did not require or imply the framing or premises which you observed were missing from my post.  A response which accounted for those would necessarily have been quite a bit longer, or would have been forced to assume a great deal of context which I am unwilling to take as given - certainly not without a negotiated examination.

I disagree with the representation that determinism and intelligence are not well-defined, in a general sense.  (They are admittedly polysemous, and some of those senses are pretty fuzzy, but as terms of art they can be quite clear.  Infinity, an analogous example, is about as well defined as a term of art as any formal object can be, and also maximally polysemous.)

A system is deterministic when its state at time t suffices to exactly infer (fully specify) its state at time t plus delta, for all t, and for all delta greater or equal to the infimum of the set of finite deltas.  (Explicitly: State need not be finite, and no Markovian property is assumed.)

Intelligence is a lattice coordinate in a partial ordering of agents, such that A <=[c] B with probability P iff P(e(A) <= e(B) | c), where e(X) is the loss of the action sequence of X given hyperparameters c.  (Interpolation over property distributions of lattice neighbors allows for a metric approximation, and dimension reduction techniques can certainly be used pack
the metric into a 1-d continuum, when context requires it.)

Then you say:  You've just shoved all the interestingness into the objective function and the definition of neighborhood.  Then I say:  An axiom schema is no less well defined than an unschematized axiom, and significantly more potent. 

Even more controversially I will propose a prototype definition of creativity as the surprisal of an observer model for the likelihood distribution of the response function of an agent.  So far, I like this because it allows one to dial-in a trade-off between formal clarity and alignment to ordinary language use of the term.  It generalizes in useful ways.

As delightful as I find Verlinde to be, I am pretty skeptical of the physicality of his theory.  There are some respected counterarguments which I have not done the dilligence to examine.  I am hoping he's very close to right, because it would confirm my own long-held metaphysical biases in a particularly gratifying way, but I am not making book on it yet.

If you expect me to prove that space is metric, I'll just shrug, since I don't even believe that it is.  Metricity is just a useful model, and there are no experiments feasible to exclude that the reality is only approximately metric.  I think that's why I balk at your ab initio calculations so often:  I am much more skeptical of the reliability of metricity assumptions.  I usually try to restrict my assumptions to weak topologies, which seem less likely to be mistaken, as they exclude fewer possibilities, and are in the main not relevantly less powerful.  In general I find myself with too much bias and not enough variance, so it is my preferred direction of model error.

First intraday support band for XMR looks like 260k-270k range.  Second support 248k, and third support 228k.  Such a wild ride!

Of course on a daily or weekly basis, it's still in a confirmed uptrend.  That won't change any time soon unless it breaks all intraday support levels, hard, with a bullet, which is a <1% chance.

My guess is that we make new 2016 highs in August, somewhere between 500k and 600k. 

You should think of public wifi as a card skimmer

 And the odds.  See 538 for that.  Something like 23:77.  But "what does it matter"?  #Rosatom will get her impeached anyhow.

Seems to me you hit the nail pretty square.  I keep squeezing him for crypto forecasts, but I haven't found a way to interpret them in a suitably consistent  way so far.  More training might help.  I should add some CNN/RNN layers in my copious spare time, maybe some sentiment and news analysis too!  Alas, life is short and full of joyful diversions.

Checked with my monkey, despite that he never seems to do well with crypto on sub-hour or daily + bars.  (Still doing quite will with FI, commodities, FX and equities though, on all scales -- kept me profitable every single day since Brexit, on hedged trading of swings in Equities, ETPs, sovereigns, and gold.)

Monkey thinks it looks grim daily right now (which could just be how a bear trap looks to the monkey), but in 2-6 weeks expects BTCUSD shooting for 975, and with some confidence.  He indicates that any serious correction or protracted sideway movement is probably 2 months away, and expects a general longer-term uptrend.

Intraday, the monkey thinks it is just beginning a little ramp up, right now, which should run a few hours.  Not sure it can break out of the 640s, in that time, but if it does, look for a continuation trend towards 677.  Current intraday support is  618.  These are coinbase prices.

On the daily chart support is 444, and resistance is 851.  STARC band is 580 to 742, daily.

The last intraday bottom was called out most clearly by the monkey when he was looking at 2 hour bars.  On that scale, he thinks we're in a downtrend likely to turn up in 14 hours or so, but he's wishy=washy and thinks it could flip to an uptrend on that same time scale at any moment.  On the one hour bars where he has also been doing pretty well lately, he thinks it looks like a 9-12 hour uptrend may be beginning now, and places resistance at 680, with 633 possible support.

I don't generally agree with the monkey on crypto, but he's pretty even-handed and usually or or another of his scenarios can be picked out as more likely using coarser indicators, stoch rsi, or such. where the 2 hour on Huo-bay-bi is starting to look like it has some serious upside potential.

If I had to pick one of his scenarios to trade, I would be going long, if I weren't already at an appropriate upside risk level on other grounds (which I am).

You can look at the June22 and July7 lower bounds as a pair of higher lows, lacking only a higher high above 4750 yuan, or 705 BFX, to be a confirmed uptrend, which wouldn't be likely to happen before July 15, in any case.

In other news, he wants me to start a small position in long GBPJPY, and to triple down in about 4 days, if it goes against me.  Probably more risk than  I want right now, though, and the fundamentals don't (yet) make sense to me, so I won't.

I am following his advice by shorting Treasuries over the next week or so, selling premium on  SPX with an August put anchor via diagonals, and just now hedging an ongoing bull gold trade.  He also has me short DAX in a small way.  I'm not sure if recent success on legacy instrument markets is attributable more to his picks or to my strike adjustments and scale-in/scale-out strategy.  I should really run more scenarios, were it not for the busy day job.

Most of my gains lately come from laddering on XMRBTC as it rises in month-long (so far) mostly steady uptrend.

Indeed, almost all crypto purchases and liquidations alike achieve interface to fiat via bitcoin exchange, even today.  It's just efficient enough so that the motivation to build a parallel system of exchanges is below threshold.  As a result, growth in alts just means more growth in BTC vs fiat.  Fortunately some alts are growing in BTC terms as well, over the long haul, so that synergies prevail.  I'm looking at you, XMR.

I've held my crypto in a Roth vehicle since 2013.  The value has increased  12x.  I saved more than 4x my original investment in tax liabilities already.  Keep in mind that if you need money, you can always borrow from the IRA.  You have to pay it back with interest, but you are just paying interest to yourself, so it's a wash, long-term.

I don't think you'll persuade anyone unless you use facts and logic.  Oh, wait, Donald Trump.  Nevermind.

That's a topic best reserved for the XMR Speculation thread here bitcointalk.org/index.php?topic=753252.new#new

I did as well, and also would consider it a fair cop if I got back 50 - which I would then double and contribute to an FFS project seeking similar ends.  Is Atrides amenable to such an outcome?

At the very least, the excercise has increased your understanding of the systems behavioural surface.  And the policy of seeking such understanding is not  unlikely to save you from loss, if only lost opportunity, in unforseeable future circumstances.

To which I would add the balancing observation that the U.S. 10 year is currently yeilding -1.5% after CPI indexing and marginal taxes.  Anybody buying treasuries today is (1) speculating on policy actions, (2) forced by a mandate, (3) gaming something linked, or (4) happy to lose less money than they would lose elsewhere.  Frankly I find it hard to put myself in their mental space.


Yeah, it's amazing the crazy stuff people do to hew to some broken self-concept, which they can't even consistently embody.  Politics does that.  I still have great respect for the practical intelligence of the German entrepreneurial class and petty bourgoisie.  Enslaving southern Europe from Slovenia to Portgual without firing a shot was one shrewd piece of work.


Here's where we are in unequivocal disagreement.  I say xauusd bottomed in January, and consider the fundamentals too dismal to allow equities to make new real highs (not nominal), and the political situation too fubared to allow even a kamikaze move like BoJ buying half the real estate in Japan to offer reflationary substance.  Blips yeah.  CPI yeah.  Pockets of growth, sure.  System down down down.


Oh absolutely.  The similarities far outweigh the differences.  It's like feeding Hokkaido uni and Santa Barbara uni to a cowboy side-by-side.  The difference just doesn't make a difference unless you're dialed into it. Even the victims of the inherent injustices are more similar than different.


...you say from a Phillipine perspective.  Japan, Korea, even China(!) not so much.  Better say South and Southeast Asia.  Then the exceptions become inconsequential.


Okay.


This bit?  Serendipity is just a stochastic algorithm.  In no wise is it the hard part of constructed intelligence.  Simulated annealling has been a staple of non-convex optimization since Metropolis-Hastings.  I know of no reason to think that any emergent phenomena of the physical substratum is not amenable to simulation to a degree of accuracy sufficient to generate every perceivable organizational behaviour required for any clearly definable pragmatically useful behaviour or outcome.  Quite the contrary:  In every case studied, the consistent pattern has been for the accuracy of simulation models to increase to within some infinitesimal delta of the principled bounds of information, where any further precision becomes an accuracy illusion, and the differences no longer make a difference.  

To claim otherwise would require that you resort to bounded rationality (as in e.g. diffusive maps bifurcating to chaos, in which case I argue that there is no evidence that chaotic divergence is germane to intelligent behaviours in a way which digitally generated chaos cannot suffice, and indeed is generally an epiphenomenon which can be bundled up in summary statistics or a lyapunov exponent, stuffed in a noise term) or hold that there is a ghost in the machine, a transcendental element, with causal import, which is masked by stochasticity.  In physcial reality, all observation is consistent with the view that, while outcome and observation is stochastic, the distribution envelope evolves with simple determinism. And non-physical reality is not amenable to rational analysis, so I consider discussing it in a rational discourse to be a red herring.  (Arational discourse has its place, but not in seeking rational clarity, with universalizable implication value.)   Omniscience is not possible (in the sense of a computable clockmaker universe, or a pervasively comprehensive sensor network, or even a complete reasoner), but neither is cognitive nihilism a useful response to the (often exceedingly well-understood, but with some subtle quirks of great interest which are less clearly understood) bounds of both analytic and synthetic knowledge.  Gödel numbers and wave equations do not constitute a mystical elan vital.  They are just approximative grammars and should not be reified.

You appear to consider that the ability to allocate new descriptive dimensions from an inexhaustible resource latent in the equations is tantamount to performing a transfinite path integral over the whole space, but it is not - no more than evolution, having solved a problem, can be said to have considered and discarded all of the alternative solutions.  (But before addressing this mistake I would point out that being latent in the equations implies that there is something to be drawn out - and I am saying it can be drawn out, in a pragmatically sufficient manner and degree for any known practical purpose, at least in principle, albeit not yet in every case has that capacity been exploited sufficently to manifestly preclude informed disagreement on its eventual outcome.)

I agree with your implied chain of reasoning from omniscience to nihil sum, and its inferential reductio, but I dispute that one crucial premise, without which the argument does not pertain.  Yes, finitary systems are incapable of embodying critical behaviours of higher order intelligence, including distinctly creative aspects.  But physical computers are not even finite closed systems, like an automaton.  They are embodied things.

Are you familiar with genetic algorithms for VLSI routing invoking flaws in the die to bypass clock propagation?  I can't quickly come up with the reference.  I think it was in one of the early 90s GA conference proceedings.  Anyhow, it wasn't an isolated case, and it seems exactly on point to the kind of limitation you would impose on automation, and to confront those limits quite directly.  And we use that sort of stochastic search power all the time, in non-parametric Bayes, in non-parametric latent dirichlet allocation, in variational autoencoders, in neural gasses.  I think you haven't chased down that rabbit hole to see how far it goes, because if you had done so, you would be saying with me:  It's rabbits all the way down.  In that case the embodied implementation manifested information in a open and unbounded form, by appearing in the objective function of a deterministic chaotic search.  How much more can be drawn out by using more modern, more sophisticated heuristic architectures, such as separated policy networks, and other virtuoso tricks which seem to proliferate faster than state university PhDs?  I think what we have in hand today is likely to be quite sufficient.

Really AlphaGo was an underappreciated watershed.  The gamespace of Go is large enough to be a fully qualified proxy for (countable) infinity.  The simple architectural trick of policy network separation, in combination with a lengthy catalog of less pivotal techniques, made the difference between a heuristic algorithm for guided search, such as is applicable in chess, and a meaningfully creative learner.  If you need to add another dimension to a learner's representation space, you just add a parameter.  It's really no big deal.  It's certainly not the stopping point of creative intelligence.

The hardest button to button in closing the loop on AI is the temporal sequence chaos which allows unbounded forward chaining inference to "jump outside of the system" in the classical Hofstadterian description, not just in toy cases, but quite generally.  It's not in the static representation space.  It's in the dynamical system.  But I have seen enough successes (and failures) in process discovery to be quite committed to the task of proving concretely that the fitness landscape is rich enough, and the requirements of practical intelligence weak enough, so that effectively unbounded arbitrarily good approximate covers of the space of intelligence behaviour ensembles are possible.   (I do not claim they are finite covers.)

It's like this:  Who really cares about transcendental numbers except as theoretical objects?  Yes, overwhelmingly all of the reals are transcendental.  But algebraic numbers, or a truncation of an infinite algebraic series, will suffice for any computation resulting in an observable.  I admit that the search landscape for dynamical systems embodying creative intelligence is not a perfect cover.  The inverse may be dense in that space.  But I also hold that the positive set is dense in that space as well.  We know how to add orthogonal dimensions to a representation space at will, I am sure you can agree.  All I ask is that you acknowledge that generalizing that to adding independent dimensions to a process space is not significantly harder, AND that the resulting behaviour manifold can be dense in the infinite-dimensional manifold containing the most precise possible description of every intelligent (creative) behaviour.  The last is not practically proven, but neither is it prohibited by first principles.  

Structure and process are duals.  Where you can create in structure, you can create in process, and vice versa. In summary, the one weird trick a Minnesota dad is using to make creative AGI is to apply that creative process to self-representation, to the self-representation of the process which embodies self.  The result is a space-filling bifurcation cascade in a Banach space of truth-functionals, which suffices to dominate every control strategy.  (I am using the words loosely because I'm not trying to do a proof here, just describe in tight summary a constructive strategy using coherent and illustrative language with sufficient clarity to make the fulfillment - with lots of sweat and capital - plausible in every regard.) I can go on about the pragmatic strategy elsewhere.  This is only to emphasize that it is not a project predicated on a contradiction.

Oh, I negated that back at the end of January.

We should have a few days' pullback or sideways here, but then it's back to up-n-up, with the monthly trend maybe taking a correction in the late fall.