Blog

Came across a situation where DevOps-style ML (deep-l ..?) was used in flash financial business. Without too much QA; merely empowering the developer/trader-trained staff to do their thing. Asked about architecture, complexity, legacy and (Taleb-style) fragility.

Got no real answer. Seems like a risky thing. Not as in risky == having some Normal distri, but as in risky == bet the business including the well-being of all employees and their extended families for a generation and a half.

As is (over?)summarised in this tweet:

We’ve gotten to the point where many companies are building faster than they can learn. Velocity is *not* the problem.

Kicker is… if you do this long enough eventually the complexity catches up to you. And velocity/ability to “deliver” IS the problem. #prodmgmt #ux #design

— John Cutler (@johncutlefish) April 29, 2019

Anyone out there that has a clue about how to Control these sort of things? I have, I can. Just call.
And:

[’cause I need a refill]

Am I or not ..?
That depends. I’m willing to go for not, though. Slacking it is.
As for the read, I’d say have a look back at this, and this (do these still exist?), and see how far we’ve come swinging back to that again, then. Per the above link. Which I enjoy.
In an unrelated note, recall this. Also, this:

I trained an AI (ML) system on a vast array of input, and found that the best way to become a billionaire is to be a child of one. Slaving away your life ('perseverance') leads to poverty throughout your life with odds a billions to a handful. Extremely irrational!

— Jurgen van der Vlugt (@jvdvlugt) July 1, 2019

That’s All, Folks!

But for:

[If only they’d bring back this, in a (technology-wise) modern version without touching the style… Then flying could be beautiful again.]

If you know what the title refers to, you have my sympathy.

Was thinking: Under (sic) the GDPR, just any EU country can have their DPA issue opinions (‘guidelines’ – you know how they go) at will, regarding clarification of apparently foggy GDPR clauses or constructs. These are then communicated EU-wide, and in the level-playing-field spirit, taken as serious advice for ‘local’ DPAs to be adopted (or face later complaints…).

This sometimes goes right, sometimes not so much.
Point is; these things get taken as authoritative (try fighting DPAs through the courts; all will be long dead economically before any conclusion would be reached) – all across the EU but those ‘local’ DPAs don’t have much jurisdiction there. Like, none. Or sovereignty has been abandoned.

Or will any DPA go as far as stating that some of their colleague’s guidance was in error, or will not be adopted, opening up the avenue for claims (non-level playing field), etc. …!?
Think again. Your thoughts, please…

And:

[Or we’ll have to sail to other shores, then… As if privvezee is such an issue there.]

An uptick I see. In the number, and profundity, of posts about ‘IT departments’. One of the more recent, and insightful, this:

With various others, older and more recent, focusing on about the same issue. It ties in with this, on the vagueness of the Value definition, perpetuated also by this.

All being Bad. But unless that Value thing is defined better and IT departments are allowed to pursue the resulting objectives on their own without ex-ante efficiency targets over innovation effectiveness targets, we’ll be stuck or be running in circles, not necessarily of the virtuous kind…

Any ideas, about that V trope ..?
[Edited to add: The above-linked methodology pusuer has more already, so track his thoughts and then add…]

Plus:

[Perfectly to spec. This was at … gee don’t recall; temporary exhibit anyway]

Would anyone have an update whether these kind of people are still onto this ..?

1. ‘These kind of people’: When displaying either such a lack of logical understanding or such a blatant urge for blabbering about (actually) ‘fake news’ a.k.a. ‘lies’, the choice is theirs and ours: Should society allow such persons to walk freely, or (xor) should they be given placement in sheltered workshops only ..? The latter, their current positions aren’t; overpaid and with the extreme negative impact they have on civil (and civilised) society at large… Yeah I kno, their deceit (if that’s what it is) is fed by other objectives – but these other objectives are against the objectives they were supposed/tasked to aim for; in case you or they don’t see that, you’re in the illogic territory again.

2. The update, on this: The discussion has been age-old, and has as long, favoured the survival of the anti-this over the pro-this, by the latter removing themselves through this, from the gene pool. But at what cost to others …?

3. Hence: We’ll have to encourage the pro-this to continue at speed, so they’re the ones being removed. E.g., by ‘counterfactual’ evidence i.e. use their backdoors into their back openings and out the results in full view. That seems to be the only way to minimise the externalities…

X: Russell’s question re Dunning-Kruger. And:

[Their brain. Or the Amersfoortse Kei. I always keep mixing these up…]

Was thinking – which at least I do, what about you ..? – on the subject of theories of firm, and on transaction costs et al. that seem to be all the rage over the last couple of decades, to be gotten rid of by ‘disruption’ and the ‘net.

Where originally, people banded together to form a firm [for the alliteration, again], so that all their individual interests could be taken just a bit further than they would have been able to on their own in sacrificing a little bit of their independence, by saving on transaction costs that would have bugged the network-of-purely-independents that was the alternative.
But at what cost? Nothing in life is free.
The coordination costs may be smaller, but when firms grow, these are back big [same] time.
The sacrifice may appear small at first, but when firms grow, these are back big [same] time.

Just re-re-re^x-read J.K. Galbraith (as referenced in here) on the four solutions. Of which a number may, might, reflect onto latter-day disruptive business models — that mostly aren’t, at least not in a sustainable way: At some point in the (near!) future, the lack of sustainable profits comes to roost. Still, read up on the (lower part of) this, for a glimpse of a better future.

On the other (?) hand; large, stable, incumbent organisations may just be too big to still have the right balance between extra productivity vis-à-vis individual goals (certainly when employees don’t see any benefit, bonus/malus, of extra effort; just coast and wing it so at least you don’t loose too much) and on the other hand feeling lost among the anonymous mass thus not seeing any personal gains. Beyond mere pay checks but there, Maslow comes in [not with a pyramid you …; read here].

Now, is there some sort of Dunbar’ish number one can ballpark refer to for an ideal, optimal organisation size ..? Where the lost and lonely are engaged back into giving just a little extra and receiving more back, outdoing the Overhead by small enough size and less internal coordination costs? Or is that where the Dunbar-150 actually comes from …??
You probably know more research than I do [0] that has been done on this already. Glad to receive your pointers…

For the TIA:

[Your org design; National Gallery, Sculpture Garden DC]

In this renewed age of privacy [attention, not actual], regulators seem not to be of this age. They still issue fines in the paltry single-digit billion order of magnitude, if at all.

Whereas this age is also one of this. Whaddif the common 4%-of-turnover were replaced by 4%-of-market-cap ..?

Would be easily implemented; take the average market cap over the last year, or over the period of breach of regulation(s), whichever is highest.
No need anymore for int’l tax reform on global dodgers, right..? B/c with the above, only ethics would remain as an argument for such reform – hah hah, in today’s age.

Why not ..?
And:

[Reality is that warped already anyway; Toronto]

Hey, just a question: Would any of you – yes, yes I know that’s a plural where a (much rounded up) single would suffice – have actual experience with working via this ..?
I posted about it some time ago [like, 39 months, oh how time flies like an arrow, like here, and here in particular] and had lost sight. But see that they seem to be quite alive, still, or am I looking at some internet artifact ..?

As a TIA for any reply…:

[Darn! Yes they do have cubicles (‘cells’ even here…) and here’s that pic again; Segovia]

No you don’t.
To be more precise, … read this. Which links back to my, and many others’, posts about modelling being all about dumbing down complication or complexity, until you ‘understand’ what’s going on. Here, taken to its full consequence.

Also, tracking back and seeing how ‘we’ may think we’re above-average intelligent, not only does Dunning-Kruger kick in big time, but also … Where do we draw the line, where is ‘our’ or others’ thinking so dumbed-down that we cannot speak of ‘intelligence’ anymore? Oh for sure, the line is somewhere between your intelligence and whatever it is that’s going on in the brains of your co-workers (let alone bosses), but how far on from there would we need to concede that machine learning is off-limits (as well) ..!?
Because it’s a scale, and a hard cut-off will fail [for the rhyme] – AI can drop the I. How are your A-systems coming along ..?

Cheers,

[Ugly, or stylish? At least, it’s consistent and it has a great bar downstairs; Amsterdam]

To throw in a new phrase. Since ‘AI’ in itself hasn’t the buzz anymore it once had, we seem to need even more interesting-looking mumbo jumbo to attract attention of the not-so-in siders.

And since ‘adversarial AI’ has been taken. In a somewhat-right way, though grossly incomplete [the doc behind it, here]. Since the most pernicious, class-breaking sub-fields have been omitted [from that linked piece]. Like, adversarial examples… Yes, yes I know I read the darn thing I know adversarial examples are in there, but not in the way that phrase used to be used (like, ages i.e., mere couple of months, ago). Diluting the term.

When ones are diluting the term for their own benefit, profit,
the conclusion is: We need a new term.
Hence, to the linked doc: Advertisial AI. Offering no solutions, just scaremongering you into contacting them so they can deliver yet another powerpoint-cum-ridiculous-billing.

… Yet again, the doc has some insights, like giving an overview of current-day pressing issues, to be tackled [note: ‘resolved’ you can’t!] before they become insurmountable.
Hey, I just want a serious piece of that action, capice..?

Edited to add: The true use with the true definition of adversarial AI, to be used against this, of course. And:

[No relevance (huh, like, the object being just a little incomplete..?), for a pretty sight; Baltimore]