Tuesday, January 22, 2019

Counting on ten left thumbs?

Software is incredibly important. It's essential to everything from managing power and communication networks, to the routine operation of vehicles and industrial processes, to the record-keeping that underlies pretty much all modern finance, to enabling blogs like this one. Which is why bad and/or over-hyped software is so problematical.

I've commented (vented? harangued? ranted?) often enough about privacy violations and security breaches in major software-based systems. Think: the OPM hack, or the compromise of every user account at Yahoo!, or the Starwood Hotel data breach. But today's post deals with other ways for software get us into trouble ....

It's too often like this. Right?
Are you one of the millions (billions?) who depends on Windows for at least part of the day? Have you noticed that not until January did we read "Windows 10 October 2018 Update is at last being pushed automatically." So what's been the delay? Try this: "Worst Windows 10 version ever? Microsoft's terrible, horrible, no good, very bad October." Which concludes with the bone-chilling prediction:

... come next April, when the 19H1 version is approaching public release, a lot of people will be holding their breath.

On to blockchain, the computing technology which underpins cryptocurrencies (such as Bitcoin). Okay, I admit it, I've blogged skeptically about cryptocurrencies, too. But blockchain is bigger than cryptocurrency. Blockchain -- at its core, a (potentially) robust, efficient, and secure method of distributed record-keeping -- could be important in its own right. But some caution is in order --

Let's look under the hood
As is spelled out in considerable detail in "Blockchain’s weakest links: Billions of dollars are riding on a technology that could revolutionize how business is done—if glitches don’t break it." Said weak links include:
  • conflating public and private blockchain (the latter of which often has vulnerabilities that go unrecognized by press coverage).
  • the sustainability of the underlying economic model, related to the viability of transaction fees.
  • the absurd level of energy consumption associated with the distributed computing model underpinning blockchain.(*)
  • regulation -- or, rather, the lack thereof. "Initial coin offerings" as a funding mechanism for startups is the Wild West of finance. 
  • And, inevitably, security issues.
(*) Consider the energy consumption of only one blockchain application: As Bitcoin prices surged, so did mining and its impact on the power grid. If Bitcoin were a country, it would rank 39th in worldwide energy usage, behind the Philippines (38th) and ahead of Austria (40th), according to Digiconomist’s Bitcoin Energy Consumption Index. Yet it facilitates fewer transactions annually than the Visa credit-card network does each day, according to Australian National University’s June Ma and Rabee Tourky and University of Toronto’s Joshua S. Gans. 

While we're on the topic of over-hyped computing technologies, of course we must talk about artificial intelligence. Real AI, not the special-purpose apps that may do one thing moderately well. The kind of AI genre readers know, love, and still wait for.

I found the following article refreshingly realistic about the challenges yet to be overcome before our silicon overlords can arrive. From author, blogger, and theoretical physicist Sabine Hossenfelder, consider "The Real Problems with Artificial Intelligence." A key assertion: 

Artificial Intelligences at first will be few and one-of-a-kind, and that’s how it will remain for a long time. It will take large groups of people and many years to build and train an AI. Copying them will not be any easier than copying a human brain. They’ll be difficult to fix once broken, because, as with the human brain, we won’t be able to separate their hardware from the software. The early ones will die quickly for reasons we will not even comprehend.

On the bright side? All that buggy, fragile, and/or over-hyped software is grist for the authorial mill :-) Except that in decent fiction, we can't have recourse to software as easy to hack, or as prone to crash, as the dreck with which we must cope every day ...

No comments: