Stupid responses to wicked problems, part [x]

Paul Raven @ 03-08-2011

Seems lots of people can see the potential long-term problems with the plans of Foxconn (and doubtless many others) to replace human manufacturing labour with robots. Sadly, that doesn’t preclude them coming up with the most myopic and reactionary response possible:

Despite my love of robots since childhood – as the high point of technology and for the technological challenges they present – we must remain vigilant about how they are helping us. If it turns out they are making our lives worse, I will be first in the luddite line with my sledgehammer.

Yes, Noel, yes! Because it’s the robots that are deciding the course of macroeconomics, isn’t it? Sneaky robots! Thank heavens for your vigilant sledgehammer; I shall sleep easier at night knowing you’re watching for that critical moment when a systemic drift manifests as an observable (if ill-defined) impact on our privileged Western lifestyles, ready and willing to destroy the tools of potential oppression, yet leaving the hands that would wield them unharmed!

Idiot. We cannot detach ourselves from our technologies; we are a cyborg species and always have been. Hairshirt back-to-basics primitivism is as unachievable and naive as Singularitarianism. Robots are tools, just like looms; why destroy a morally neutral tool when you could instead work on the systemic problems which make that tool into a vector of oppression?

Fight the fist, not the gauntlet.


Wicked Problems and ends to limitless [x]

Paul Raven @ 02-08-2011

That Steelweaver post on Reality As A Failed State I mentioned a few days back really did the rounds. So I’m going to link to Karl Schroeder at Charlie Stross’s blog once again, and without any sense of shame – he’s been quiet for ages, but he’s spooling out a year’s worth of good shizzle over the space of a few weeks at the moment, and I think he’s a voice worth paying attention to.

Here he is talking about the “metaproblems” that Steelweaver mentioned, which have not only been known and named (as “wicked problems” for some time, but are already a subject of intense study… which is a good thing, too.

It is not the case that wicked problems are simply problems that have been incompletely analyzed; there really is no ‘right’ formulation and no ‘right’ answer. These are problems that cannot be engineered. The anger of many of my acquaintances seems to stem from the erroneous perception that they could be solved this way, if only those damned republicans/democrats/liberals/conservatives/tree-huggers/industrialists/true believers/denialists didn’t keep muddying the waters. Because many people aren’t aware that there are wicked problems, they experience the failure to solve major complex world issues as the failure of some particular group to understand ‘the real situation.’ But they’re not going to do that, and granted that they won’t, the solutions you work on have to incorporate their points-of-view as well as your own, or they’re non-starters. This, of course, is mind-bogglingly difficult.

Our most important problems are wicked problems. Luckily, social scientists have been studying this sort of mess since, well, since 1970. Techniques exist that will allow moderately-sized groups with widely divergent agendas and points of view to work together to solve highly complex problems. (The U.S. Congress apparently doesn’t use them.) Structured Dialogic Design is one such methodology. Scaling SDD sessions to groups larger than 50 to 70 people at a time has proven difficult–but the fact that it and similar methods exist at all should give us hope.

Here are a few wicked problems I think are exemplary. I touched on one of them yesterday, in fact, namely the roboticisation curve in manufacturing; far from liberating the toiling masses in some utopian fusion of Marx and capitalism, it might well increase the polarisation and widen the gap between the poor masses and the super-rich elites, a process that Global Dashboard‘s Alex Evans refers to as “jobless growth”::

In some developed economies (and especially the US), research suggests that job opportunities are increasingly being polarised into high and low skill jobs, while middle class jobs are disappearing due to “automation of routine work and, to a smaller extent, the international integration of labour markets through trade and, more recently, offshoring”. Meanwhile, data also show that while more women are entering the global labour force, the ‘gender gap’ on income and quality of work is widening between women and men. These trends raise a number of critical uncertainties for employment and development to 2020.

If automation of routine work genuinely is a more significant factor in developed economy job polarization than international trade or offshoring, then the implication is that developing economies may increasingly also fall prey to job polarisation as new technologies emerge and become competitive with human labour between now and 2020. Chinese manufacturing and Indian service industry jobs could increasingly be replaced by technology, for example, and find their existing rates of inequality exacerbated still  further.

And here’s a serendipitous look at the economics of a world where replicators and 3d printing become cheap enough to be ubiquitous [via SlashDot]:

Prices for 3D printers are tumbling. Even simple systems often cost tens of thousands of dollars a decade ago. Now, 3D printers for hobbyists can be had for a fraction of that: MakerBot Industries offers a fully assembled Thing-O-Matic printer for just $2,500, and kits for building RepRap printers have sold for $500. The devices could be on track for mass-production as home appliances within just a few years.

So, will we all soon be living like Arabian Nights sultans with a 3D printing genie ready to grant our every wish? Could economies as we know them even survive in such a world, where the theoretically infinite supply of any good should drive its value toward zero?

The precise limitations of replicator technology will determine where scarcity and foundations for value will remain. 3D printers need processed materials as inputs. Those materials and all the labor required to mine, grow, synthesize or process them into existence will still be needed, along with the transportation costs to bring them to the printers. The energy to run a replicator might be another limiting factor, as would be time (would you spend three days replicating a toaster if you could have one delivered to your home in an hour)? Replicators will also need inputs to tell them how to make specific objects, so the programming and design efforts will still have value.

[…]

Perhaps the most important limitation on the replicator economy may competition from good old mass production. Custom-tailored suits may be objectively better than off-the-rack outfits, but people find that the latter are usually the more sensible, affordable purchase. Mass production—especially by factories adopting nimble 3D-printing technologies—can still provide marvelous economies of scale. So even when it is theoretically possible for anyone to fabricate anything, people might still choose to restrict their replicating to certain goods—and to continue making their tea with a store-bought teabag.

The unspoken underpinning of that last paragraph (as hinted by my bolding) is the important bit: the economies of scale of fabbing will see more and more human labour replaced by machines – machines that don’t need holidays, or even sleep; machines that don’t get tired and make a higher percentage of dud iterations as a result; machines that, before too long, will be able to make other machines as required. The attraction of such a system to Big Capital (and small capital, too) is pretty obvious.

And all in the name of chasing perpetual infinite growth, a central assumption of most modern economic thought (or at least the stuff I’ve encountered so far) that relies on a lot of other assumptions… like, say, the assumption that we’ll always be able to either produce more energy, or use the amount we have available more efficiently [via MetaFilter]:

It seems clear that we could, in principle, rely on efficiency alone to allow continued economic growth even given a no-growth raw energy future (as is inevitable). The idea is simple. Each year, efficiency improvements allow us to drive further, light more homes, manufacture more goods than the year before—all on a fixed energy income. Fortunately, market forces favor greater efficiency, so that we have enjoyed the fruits of a constant drum-beat toward higher efficiency over time. To the extent that we could continue this trick forever, we could maintain economic growth indefinitely, and all the institutions that are built around it: investment, loans, banks, etc.

But how many times can we pull a rabbit out of the efficiency hat? Barring perpetual motion machines (fantasy) and heat pumps (real; discussed below), we must always settle for an efficiency less than 100%. This puts a bound on how much gain we might expect to accomplish. For instance, if some device starts out at 50% efficiency, there is no way to squeeze more than a factor of two out of its performance.

[…]

Given that two-thirds of our energy resource is burned in heat engines, and that these cannot improve much more than a factor of two, more significant gains elsewhere are diminished in value. For instance, replacing the 10% of our energy budget spent on direct heat (e.g., in furnaces and hot water heaters) with heat pumps operating at their maximum theoretical efficiency effectively replaces a 10% expenditure with a 1% expenditure. A factor of ten sounds like a fantastic improvement, but the overall efficiency improvement in society is only 9%. Likewise with light bulb replacement: large gains in a small sector. We should still pursue these efficiency improvements with vigor, but we should not expect this gift to provide a form of unlimited growth.

On balance, the most we might expect to achieve is a factor of two net efficiency increase before theoretical limits and engineering realities clamp down. At the present 1% overall rate, this means we might expect to run out of gain this century.  Some might quibble about whether the factor of two is too pessimistic, and might prefer a factor of 3 or even 4 efficiency gain.  Such modifications may change the timescale of saturation, but not the ultimate result.

So it ain’t just Moore’s Law that could be running into a brick wall real soon. A whole lot of caltrops on the highway to the future, then… and we’re still arguing about how to bolt more governers and feedback loops onto fundamentally broken polticoeconomic systems. Wicked problems, indeed. It’s hard not to feel bleak as we look into the eye of this abyss, but Schroeder suggests there’s a way out:

Here’s my take on things: our biggest challenges are no longer technological. They are issues of communication, coordination, and cooperation. These are, for the most part, well-studied problems that are not wicked. The methodologies that solve them need to be scaled up from the small-group settings where they currently work well, and injected into the DNA of our society–or, at least, built into our default modes of using the internet. They then can be used to tackle the wicked problems.

What we need, in other words, is a Facebook for collaborative decision-making: an app built to compensate for the most egregious cognitive biases and behaviours that derail us when we get together to think in groups. Decision-support, stakeholder analysis, bias filtering, collaborative scratch-pads and, most importantly, mechanisms to extract commitments to action from those that use these tools. I have zero interest in yet another open-source copy of a commercial application, and zero interest in yet another Tetris game for Android. But a Wikipedia’s worth of work on this stuff could transform the world.

Digital direct democracy, in other words, with mechanisms built in to ameliorate the broken bits of our psychology. Oh, sure, you can scoff and say it’ll never work, but even a flimsy-looking boat starts looking like it’s worth a shot when the tired old paddle-steamer starts doing its Titanic impersonation in the middle of the swamp. What Schroeder (and many others) are suggesting is eminently possible; all we lack is the political will to build it.

And it’s increasingly plain that we’re not going to find that will in the bickering halls of the incumbent system; it’s only interested in maintaining its own existence for as long as possible, and damn the consequences.

Which is why we need to turn our backs on that system and build its replacement ourselves.


Foxconn’s robot recruitment drive: the beginning of the end of labour?

Paul Raven @ 01-08-2011

Alex Knapp picks up on the story about Foxconn’s plans to draft in a cool million robots for their manufacturing plants within three years, and ponders whether this is good news for the unemployed in the West:

A good portion of this move to robotics labor has, no doubt, to do with labor costs and the reports of working conditions in FoxConn’s factories in China. I do wonder, however, what this means for the future of outsourcing to Asian markets, though. If labor costs can be reduced by employing more robots in the factory, and if its feasible for a company like FoxConn to use such a large number, it begs the question of why Western companies might continue outsourcing.

One of the major reasons, after all, that companies are outsourcing their operations to Asia is for the labor costs. If those labor costs can be obviated by greater automation, then other considerations come into play. After all, America and Europe are still where most of the customers are, and with the price of oil on the rise, cheaper transportation costs might come into the mix. Moreover, with both Europe and the United States teetering on the brink of another double-dip recession, policies geared towards bringing manufacturing home via tax incentives and other measures are more likely to become law.

Hmm. Even allowing for the traditional hyperbole of corporate press releases, a million automated machines will displace a whole lot of jobs. But Knapp is far more confident than I am about governmental willingness to pass laws that make demands of their corporate paymasters, let alone ones they’d actually be obliged to adhere to; the geographically-bound nation-state has no leverage on the transnational corporation, and the only incentive you can offer them is a taxation low-ball and further erosion of worker’s rights. In other words, sure, you might get manufacturing returning to the West, but it’ll only happen because labour costs have equalised… and that’s only going to happen in places where you’ve carefully created a huge stock of labour that doesn’t have any other options. Sweatshop USA… it’s a sign of how bad things are if you can count that as a victory condition.

And to extrapolate further, the rapid maturing of fabrication technology is going to obsolete a lot of more complex machinist gigs, too. They may well be countering fabbers and sinterers among the Foxconn “robot” list, of course, though the article mentions more simple stuff – welding, spraying, low-level assembly, which are exactly the sorts of jobs that robots have traditionally obsoleted before. I watched pick’n’place machines and surface mount technology completely gut the electronics assembly industry here in the UK back in the nineties, and less than a decade ago it was already cheaper for one company I know of to source components internationally, ship them to the UK for collation, ship them out to Thailand for assembly and then back again for QA and (extensive) rework, than it was to pay UK workers to do the full process locally. That sort of price margin isn’t going to be significantly eroded by concessionary tax rates, though increased transportation costs will admittedly contribute as well.

I think if there’s anything to take away from this inexorable slide toward nigh-total automation, it’s that we can no longer sustain a global economy that relies on mass employment in the manufacturing industry to keep the money moving. I suspect there’s a tipping point in the near future (or possibly in the recent past) at which – barring existential-scale disasters that knock us back into pre-industrial ways of life – it will never again be cheaper to get humans to build things than it is to get machines to do the same work.

What happens after that point to the business models of tchotchke makers and Next Big Thing gadget creation, I have no idea… and the more economists and politicians I listen to, the more I suspect that no one else has a bloody clue either. If anyone can show me how this *isn’t* a big old zero-sum game that’s going to hit a brick wall sometime soon, please pipe up and do so.


US and China to have manufacturing costs parity by 2015?

Paul Raven @ 09-05-2011

I’m going to offer this with a large pinch of salt, given that it’s a press release from a consulting firm, but the boldness of the claim is pretty impressive [via NextBigFuture]:

Within the next five years, the United States is expected to experience a manufacturing renaissance as the wage gap with China shrinks and certain U.S. states become some of the cheapest locations for manufacturing in the developed world, according to a new analysis by The Boston Consulting Group (BCG).

[…]

After adjustments are made to account for American workers’ relatively higher productivity, wage rates in Chinese cities such as Shanghai and Tianjin are expected to be about only 30 percent cheaper than rates in low-cost U.S. states. And since wage rates account for 20 to 30 percent of a product’s total cost, manufacturing in China will be only 10 to 15 percent cheaper than in the U.S.—even before inventory and shipping costs are considered. After those costs are factored in, the total cost advantage will drop to single digits or be erased entirely, Sirkin said.

Products that require less labor and are churned out in modest volumes, such as household appliances and construction equipment, are most likely to shift to U.S. production. Goods that are labor-intensive and produced in high volumes, such as textiles, apparel, and TVs, will likely continue to be made overseas.

Talk about a mixed bag of news. The prospect of working-class jobs returning to American shores must be something of a relief, but implicit in that return is the socioeconomic status of those “certain U.S. states” (and I think we can all guess which ones) as equivalent with China, the great economic enemy and exemplar of all things unAmerican. And it puts the lie to the notion of the unity of the US, too; sure, the top 1% of the country is rolling in money, but the bottom layer of the population pyramid is competing with China for the chance to make tchotchkes. Kinda puts the whole “USA! USA!” chanting from last week into perspective, doesn’t it? If this is a victory condition, I’d hate to be losing the game. (Note use of sarcasm as a way to blunt the pain; things over here on Airstrip One are looking grimmer by the day, too.)

Also implicit in the consultant’s outlook there is that the methodology of manufacture will remain essentially the same. Four years doesn’t look like a long time, but things move fast these days, and the 3D printing and fabbing industry is edging closer and closer to the point where it becomes a big grenade in the labour punchbowl. Still, I guess someone’s gonna have to make the 3d printers… up until the point where they can reliably self-replicate, anyway. (Shorter version: economics of mass production looking pretty screwed in the long term with respect to job creation. Profitability looking much better, but the 0.01% of the population who’ll benefit from it don’t need me to tell them that, I expect.)


Hyperlocal manufacturing: fabrication factory in a shipping container

Paul Raven @ 17-08-2010

It’s amazing what you can cram into a shipping container: a solar power generator, an internet cafe, a data centre… or a self-contained tooling workshop and fabrication unit [via GlobalGuerrillas – beware dodgy pop-ups on the linked page]:

The MPH was developed when the army realized that the easiest way to get the many rarely requested, but vital, replacement parts to the troops, was to manufacture the parts in the combat zone. In short order, this led to the construction of a portable parts fabrication system, called MPH, that fit into a standard 8x8x20 foot shipping container. The original version used two containers, but smaller equipment and more powerful computers eventually made it possible to use one container.

As John Robb and others have pointed out, this is a blueprint for hyperlocal manufacturing… though to make it economically practical you’re going to have to shave down the construction costs from the bloated levels of military contracting:

There are four MPH systems in service, two of them in Afghanistan. A fourth is being built, at a cost of $1.5 million.

Ouch. What do you actually need? If you’re going local, you just need a space the same size as a shipping container; should be a stuffed animal that’ll do the job. A fast broadband connection will stand in for the military satellite link (assuming you’re operating in an urban area); speed probably isn’t too crucial with non-military applications, so you might be able to cantenna yourself into a convenient local wireless node for big savings. Then you need CNC machines and raw materials; the former can probably be bought up pretty cheap from bankrupt stock (hell, you might not even need to move the kit if it’s still sat in a disused factory unit – two birds, one stone), and the latter scraped up from salvage and reclamation…

Anyone fancy running the numbers on this?


Next Page »