Tag: AI

Assurance… No; continuous blockchainproofing will be

Accountants (of the certifying kind) have seen the light of continuous assurance coming. The vast majority of them reacted by being the rabbits [certainly not of the Winnebago / Native American trickster type ..!]; though assuming the headlights were and are still very distant, sitting quite still…
A select few have responded differently – embracing some change as inevitable, researching how Continuous Assurance might be, in times of proliferating XBRL and the like.

That’s OK. And laudable for the Virtue of facing the danger not ducking.

But … all of the assurance industry is still lock, stock and barrel dependent on being the Third Party in agency models.
And now, blockchain tech is around the corner, promising all sorts of unbelievable new ways of transferring trust. If only one could build some system(ic) in which any principal would be able to Read all minute transactions of an agent, and would be able to reliably (…) make sense of it – then the information quality (read: [non]uncertainty, [non] information (access, processing capability) difference) would be immediately visible and actionable. Undoing the need for a trusted third party to give a second opinion that is so beaten down to platitudes anyway that the usefulness has deteriorated way beyond what third parties themselves still believe (if they wouldn’t, who would…?). And note the italics of trusted.

Trusted – the thing that blockchain technology spreads so evenly, so extremely to the opposite of the ultimate non-spread of one person/entity.

Oh well. You know now, and this:
DSC_0235
[Relevant if you think it through: Warped reflections. NY of course]

Still valid; MIS is a Mirage

Somehow, some neurons fired that sublimed into a thought about John Dearden’s MIS is a Mirage, of 1972 … Turns out I’m not the only one who thinks it’s still very much valid today. As e.g., here. Oh, the insights that run in deep undercurrents throughout today’s management- and other fads…!

But, once ASI comes along … Then at last, MIS can do without the, then relative but still, stupidity of mankind. Or ..?

[No pic today. Post too short. First, you study the article at length!]

Schmobol

With the current uptick in interest in the ageing population … of the handful still capable of hardcore manual programming of COBOL, as e.g., here, I wondered:

  • Is the code base still so enormously biased to COBOL(-based! software)?
     
  • Why haven’t COBOL-to-e.g., C, or others converters, not caught on widely so the ‘problem’ doesn’t exist anymore ..?

Especially the latter; especially since we still have tons, too many (?), programmers available to tackle C- and more modern-language scrutiny and optimisation jobs. Jobs; high-pay jobs. And automatic testers to compare absolute 1-to-1 identity of the functionality or tractable lists of boundary conditions (possibly differing).
But with so many more (modern) code/functionality maintenance tools and capabilities available. And with integration/migration to (even) newer integrated platforms available.
And, when things get tough, AI that should be easily trainable to get to the hard, core bugs (higher abstraction sense) before/after the translation(s).

So, what’s the deal? The only deal there is, is (and was, having lost a long time) the lack of forward-looking maintenance to have already started early on modernisation. Yes, of course, there’s Not On My Watch and Après Nous Le Déluge. But real leaders would cut through that; that’s what distinguishes them from mere shopkeeper ‘managers’.

All right. Leaving you with:
DSCN3028
[Impossible to guess I guess. Where?]

The need for a new security framework

… I feel the need for it. A new security framework.

Because what we have, is based on outdated models. Of security. Of organisations. Of how the world turns.
Bureaucracy doesn’t cut it no more. The very idea of hierarchically stacked framework sets (COSO/CObIT/ISO27k1:2013/…) likewise, is stale.
And the bottom-up frameworks en vogue, e.g., OSSTMM (if you don’t know what that is all (sic) about, go in shame and find out!) and core work like Vicente Aceituno Canal’s, haven’t found traction enough yet, nor are they integrated soundly enough (yet!!) into further bottom-up overarching approaches. Ditching the word ‘framework’ as that is tainted.

But what then? At least, OSSTMM. And physical security. And SMAC. And IoT. And Privacy (European style, full 100.0%, mandatory). And business-organising disruption, exploded labour markets, geopolitics, et al.

OK. Who of you has pointers to such an Utopia ..? [Dystopian angles intended]

Unrelated:
DSCN6146
[Your guess. Not Nancy. But is it Reims ..?]

Road rage (autonomous car edition)

Two things about self-driving cars:

  • When a fully ready car leaves the factory, it should be programmed completely, at least with all needed to function independently from thereon. Does it think at that very moment “Ugh here I am, just born and already with a huge traffic jam behind me (in the factory)”..?
  • Where humanoids will in school and around that, learn quite some data and algorithms by heart just to know them, and acquire experience on how to deploy all that, in the chaos that is the actual world, before being considered an adult capable of independence and accountability in the free world, how will autonomous cars gain such experience once they leave the factory ..? Will they be utterly clumsy during the first few miles / years (sic) ..? Your legal department may need to know.
  • [Third, because rebel.] News broke that self-driving cars drive like your granny. Get out the bull bars (originally: ‘roo bars! from down unda) and shove them off the road. [Edited because apparently necessary to add: In no way literally or even close, you m.r.n!]

So. Moving on ….:
DSCN4611
[Skewed before screwed; Madrid for no reason whatsoever]

Singularity / M-jumps

Musing with the ideas, suddenly (?) resurfacing the last couple of weeks, of memes being the abstract ideas that spread over human brains like viruses do in the physical world.
Where ‘virus’ taken in a wide sense may include the mitochondria et al., would reflect into the abstract like algorithms and/or Turing engines / data streams.

But that’s just some analogy to just track back. My concern was (is?) with what would happen at the point where ‘machines’ would become so intelligent, or the physical substrate that information (and/or algorithms, analogy from the other side?) rides on, would no longer need human brains, human flesh and bones, to function, procreate and spread. Is that at, or past, or before the Singularity ..? I see [wanted to write ‘envision’ but one should eschew obfuscation!] various scenarios following from there various scenarios. Not all, too happy. Let alone for us humans.
And how would we call such points? Ladies and gentlemen, I coin ‘M-jumps’.

Your comments are welcomed. Even if you expected “you’re”.
Oh well, …:
20150517_212545
[Just an off the cuff phone snapshot; Baltimore by night]

Musk / Vicarious / ASI

Haven’t heard in a while of anything fruitful coming forward from Elon Musk’s investments in Vicarious despite his concerns. (Now that DeepMind has gone over to the Dark Side.)

Reasons ..? Double secrets?
I’d just like to know.

For the weekend:
DSCN8447
[Algorithmic justice prevails ..? Amsterdam]

FogAI picture

… Just to put it out there: What has happened to all the thrilling AI initiatives that flew around one after the other at the start of the year ..?
At that time, I even included some stuff in my Predictions, as so many new things were popping up. But now, … not so much. Because what?

Or have all the ‘leaks’ been thumbplugged and is development still going strong in skunk works towards a renaissance explosion sometime soon ..?

Whatev’; for you:
DSCN2101
[Its back being Mont serrat. Or so. ?]

Signalling healthy process

Yet some more cross-over ideas from the IoT world into the administrative bureaucratic office world: Streams of transactions as signals.
Of the health of the process, of course. To be defined, obviously, as the fit to the surroundings. The fit may be off, either intentionally (wanting to let the world adapt to the process, enforcing (?) change) or unintentionally left blank                i.e., having to cope with exceptions to what was envisaged as transactions’ content or form.

Now apply yesterday’s first picture of process control.
Now, too, consider what one could do with sampling theory (as a subset of ‘Shannon’, if properly elaborated, possibly skirting with ‘classical’ statistics ..?). Taking 2log(n) samples (where n is the number of transactions ..?? Just a wild guess) and being able to reconstruct the ‘signal’ then taking its integral (discrete transactions … just summing it up ..?) for the total. Or Fourier-transforming it all and … get your basic theory straight before dreaming of moving on so don’t start at the other end as ‘accountant’…! And/or treating exceptions (as e.g., found by the sort of analysis that these girls/guys are so good at; that not even being meant as a cynical qualifier) as noise to the signal. Never fully suppressable, but useful to pick up secondary signals, stacked in their variation of frequencies, amplitudes an wavelet transformations. That all tell you something, if you listen. Whether you want perfect, over-HiFi replay [intermission: Ugh I’m getting old, even knowing that HiFi was a thing…], or lively veracity, actual fullness of music. And take in again the ole’ industrial process control with its recipe / derivative function(s), et al., and be able to better control it all from the ‘dashboard’ in the control room. When all of the routine stuff, the routine 80%, of business is done by … ‘robots’. Humanoid or digital-machines, IDC.

And hey, while we’re at it, why not throw in attempts to include in bookkeeping not only discrete numbers (arbitrarily rounded to hunderds, of random currencies) but Real numbers or even Complex numbers as well ..? The latter, e.g., to indicate VAT surcharges, etc.; leading to tuples-as-single-‘numbers’ in bookkeeping. Maybe somewhat harder to track that all is booked correctly, but also maybe powerful in capturing singular transactions and some processing rules/logic, and controls, in one tuple (‘record’).

Where AI may then be applied to do sanity checks. Not on this author; no AGI or ASI would suffice…

OK, for now:
DSCN1436
[“What a shoe box” but yes that *is* the Bata shoe museum, Toronto]

Maverisk / Étoiles du Nord