Bulletin February 23 2018. The bittersweet tragedy of falling thresholds

You have to feel sorry for Charles Babbage, whose 1830’s designs for the analytics engine were never to be realised in his own lifetime. As well as Augusta Ada King, Countess of Lovelace, the maverick mother of programming, whose own book was never finished. It could be said they were the founders of steampunk, in which all matters tekno-logical could be enabled with clever use of cogs and lenses. 

Arguably, they were also the first to experience the falling threshold theory of computation. While it was potentially possible to realising his designs with the accuracy required, the costs were not seen as acceptable by the UK Treasury. How different might history have been, had they been funded. Indeed, another hundred years passed before others attempted to devise contraptions that could perform calculations according pre-‘programmed’ instructions. 

When figures (such as German engineer Konrad Zuse, with an adaptation of the telegraph) finally did, they no doubt went through similar epiphanies of invention, those moments where the world appears to open up. “Just what I will be able to do with <insert device, program or concept here> beggars belief,” goes the thinking, repeating a pattern that has no doubt continued since the invention of the wheel. 

(As an aside, I am reminded of one of my favourite cartoons from the Wizard of Id. “I have it, I have it! The greatest invention since the wheel!” announces the Wizard. “What is it?” replies the King, to which the Wizard says, “The axle!”)

No doubt the inventors of the tin can envisaged a future in which any foodstuff could be ‘canned’… these waves of invention are so often followed by the equally inevitable realisation that, no, not all problems can be solved — not cost-effectively, at least, and not at the time. Only the advent of Moore’s Law, variants of which still continue, offer any hope: if something isn’t possible yet, then it may be so in a couple of years’ time. Or so. 

The trick about Moore’s Law is that it became an impetus to research, a budgetary variant of Heisenberg’s Uncertainty Principle, in that measuring something influences its behaviour. In this case, funds were allocated and priorities set in order to ensure it remained true. This is without malice or corruption: the potential consequences of delivering on the Law, and the opportunities thus created, were sufficient to drive the strategies of Intel, IBM et al. 

Visionaries and indeed, marketers have tried to drive similar self-fulfilling prophecy through the power of their aspiration. When this is both blatant and unlikely to succeed, we refer to this as Kool-aid, illustrative of how easy it is to get whipped up into a frenzy of belief about the next big thing. It is similarly simple to be cynical, to point out how unlikely the vision is in practice. As I have written before, we may be right, for a while. 

When breakthroughs tend to happen, they are driven by cost effectiveness, limited in scope and broad in impact: so, for example, the arrival of Hadoop onto the data management scene became possible due to the availability of certain components, notably in-memory data stores (which are lost if the machine is turned off, but that doesn’t matter if you already have your answer). Similar, the wave of “AirBnB doesn’t own a hotel”-type pronouncements. 

Most of the major innovations we have seen in recent years are a consequence of these falling thresholds. Computation really is just maths, and infrastructure really is just engineering: anyone preaching the mantra of agility is on firm ground, as the chances are very high that what we see today as the norm (and build our businesses around) will tomorrow become superseded. 

Equally, some of that maths will still be held back by the current state of engineering. Artificial Intelligence, the Internet of Things, Big Data Analytics, Cloud Computing are all models operating under various levels of constraint. The trick is to avoid listening to what might be possible, which may or may not be noise, while being ready to take advantage of the very real opportunities when they emerge. 

Innovation in its truest sense — that is, constant renewal of thinking and practice — is the name of the game. Keeping this in mind, here’s some articles for this week. 

What’s missing from the Malicious Use of Artificial Intelligence report?

While there’s a lot in this meaty report about what might go wrong about AI, it lacks substance — partially fair given that we are talking about the future, but equally unacceptable given our shared goal of understanding, and mitigating, potential risks. OK, I’m giving away the plot but the consequence is decidedly unscientific, which is surprising given that it comes largely from academia. 

Why AI on a chip is the start of the next IT explosion

Following on from the above, it’s ironic that I should appear to be writingwith Kool-aid in my veins, with that header! The bottom line is that chip design is moving from something everybody did, back in the day (my first job was Philips Components, and who remembers the Transputer?), via a phase of commoditisation, to something everybody does once again. This shift from proprietary to commodity to open, seen so clearly in software, is now happening at the chip level and will only expand.


Next week: DevOps in the Real World

What’s that, DevOps isn’t for everyone you say? As reported a couple of weeks ago, my virtual fireside chat with Nigel Kersten of Puppet Labs will be on 27th February 11.00AM Central, 17.00 GMT. If you’re interested you can sign up here

Five questions for…

Okay, this isn’t an article but a plan. Conscious of the fact that briefings with technology companies are important, and that nobody wants to waste anybody’s time, I’m starting a series of posts where I will ask five questions of a technology vendor, or indeed anyone who can help elucidate what’s going on in tech. Let me know if you are interested and we can set something up. 


Extra-curricular — a shout-out for parkrun

Finally, last Saturday I was Race Director for the third outing of the UK’s 500th parkrun venue, at the Royal Agricultural University in Cirencester. I can’t extol the virtues of parkrun enough — it is incredibly inclusive, with sprinters, walkers and people pushing prams sharing the same route. A run not a race, the goal is to get the world fitter. Long may it continue. 

Finally, thank you to all my subscribers — I’ll be sharing some stats soon. Any questions or feedback, let me know.

Until next time, Jon

Bulletin February 23 2018. The bittersweet tragedy of falling thresholds

Leave a Reply

Your email address will not be published. Required fields are marked *