Monday, September 15, 2008

Force Multipliers and The Impossible Singularity

Engineers have a different definition of "impossible" than most other people.

See when a non-engineer says something is "impossible", what they generally really mean is either "I really don't want to do that", or "it's really really hard and we don't know how to figure it out yet but we probably will in the future".

When an engineer says something is impossible what he means is "There is no known way of doing this, nor under current scientific understanding CAN there be any way of doing this".

That's a pretty significant difference.

Oh and in case you're interested, when scientists say the word impossible, what they really mean is "It may theoretically be possible if we're wrong about the science... and we may be wrong about the science but I don't think so".

Actually, most of the time engineers and scientists will qualify the word impossible by saying "that's impossible based on current understanding" or something like that, because indeed, anything is possible... just highly improbable (parents, talk to your children about quantum physics, before they get it from the street).

There is a somewhat popular futurist notion going around right now, that basically says that because of the vast increases in human knowledge and available computing power, we will reach a technological "singularity", which will push us into a post scarcity economy, because once computers are smart enough to reproduce, and improve themselves into even smarter computers, all material problems will somehow be solved.

These utopians (and some dystopians) believe this singularity will solve all of our current material and economic problems... but it will create a whole new set of problems because today economies are based on scarcity, and we have no idea how to operate a post scarcity economy.

Which is true, we don't, but there's a much bigger problem with the idea of the singularity in and of itself.

The fundamental conceit of the unconstrained intellect, is that intellect can solve any problems. Of course, intellect is NOT unconstrained, no matter how we may wish it to be so... but it often looks like it to a lot of people.

Simply put, the singularity, as envisioned is impossible, by engineers terms.

A few days ago, Eric Raymond wrote about the Vanishing Consumption Gap, and some of the details and implications surrounding that.

Some of his commenters suggested that this was yet more indication of the singularity, and the transition to a post scarcity economy (unsurprisingly Eric disagreed).

One in particular used a common example of the argumentum continuum logical fallacy; presuming progress which had been made before, would continue at the same, or faster pace:
"My sister is a PhD bio researcher. In the 80s, using stone knives and bear skins, she perfected techniques for extracting and coding fish and insect DNA. 20 years later, those ground breaking techniques that she invented are not even considered. Computerized analyzers can do more work in 10 minutes than she could in a year. Once the software and hardware was developed, anyone with a high school bio education (or very smart) can run them and interpret the results. The equipment is not cheap (5K - 50K depending on various factors), but one machine and one technician costs less than her college education did.

The smart people, or the determined, dedicated, driven people, will invent new things and get paid (in coin or praise) for it, because they must. The rest of us will use it without thinking about it or even having a clue about it.

There will always be things that are expensive, that only the wealthy can afford, but what those items are will change - and the rate of change is accelerating."
To an extent, Dan is correct; though the implications are not nearly as far reaching as one would hope.

What Dan has described here is called a “force multiplier” a.k.a. productivity enhancement, which is part of what Eric was addressing in the first post (linked above).

The force multiplier effect is extremely well understood; it’s also well understood where it breaks down.

A (slightly dramatic) example:

Currently, given anything better than 6 to 1 odds against a third generation or earlier opponent (American forces are currently fourth generation warfighters; aka post industrial or information warfighters. Iraqis are third generation, industrial warfighters. Afghanis are second generation pre-industrial warfighters - not this is not related to the "4th generation warfare" concept); an American combat units advantages in morale, training, equipment, combat support, logistics, and what used to be called C2 (command and control), and is now often referred to as C4i (command, control, communications, computing and intelligence); allow commanders to remain confident in overall mission success.

Those are all force multipliers; i.e. factors that increase the effective productivity of an individual worker; or in this case warfighter.

Force multipliers have always existed, and always will. The English experienced an effective 5 to 1 force multiplier at Crecy, with the muddy terrain and the Longbow. The Spartans had an effective 25 to 1 force multiplier at thermopylae (3300 men including all the allied forces of Greece - not just 300 spartans - held off at least 80,000 Persians) with the terrain.

The problem lies in where those factors don’t work properly, are countered by the enemy (either by possessing similar factors themselves, or by using countermeasures), or are simply ineffective; as happened in Somalia, or as might happen in human wave attacks in bad weather (the north Korean scenario)… or as happened at Thermopylae when the Persians found a way around the hot gates.

Without force multipliers, the standard Clauswitzian maxim of needing a 2:1 advantage to have justified confidence in victory, and a 3:1 advantage to be assured of victory, still applies (3:1/6:1 if facing well trained troops in prepared defensive positions).

Also note, that force multipliers are far less effective on an individual warfighter basis as opposed to a full unit basis; because not all of them apply to the challenge of one man fighting another, and because size is an effective multiplier on it’s own (presuming a units C4i and logistics are adequate).

So, while a brigade of American troops in the open field, and with the initiative, may be as effective as six brigades of Iraqis; you still wouldn’t want to enter a standup fight on those odds unless you absolutely had to; and any individual American warfighter may only be as effective as any three, two, or even one Iraqi… or under certain circumstances less than one.

Now, as I said, this is a very dramatic example, with life and death stakes; but the principle of force multipliers also applies to industrial efforts.

Dans sister had the multiplier of being extremely intelligent and innovative, and did work that 50 less educated, intelligent, and innovative people could not do at the time. Today, a single one of those less educated, intelligent, and innovative people can do 50 times what she was able to do then…

…When those multipliers are in effect...

The reciprocal is also true however. Dans sister cannot now do 50 times what the intern can do; unless she can innovate again to give herself a new force multiplier. Therefore her labor efforts have been equalized with those of a lower value worker, in this instance.

Her force multiplier is still her intelligent, educated, and innovative brain however. When she can apply that, she will be many times more effective than someone without that advantage.

...When she can apply it...

The least constrained field of engineering is probably software development; because other than the availability of raw computing resources, software is essentially unconstrained. Whatever one can imagine, one can program (getting it to run fast enough to be useful... or to have useful input to produce useful output, is a different story of course).

Now, when it comes to a software development project, we all know that some engineers are 50 to 1 force multipliers, and some actually negate others (I've never heard the term force divisor, but it would apply to some engineers I've known); but that over time, as more effective techniques are developed and disseminated, the lowest performers still become more productive, along with the highest performers.

In fact the gap in performance tends to narrow, because the lower performers are able to take advantage of techniques that higher performers were using previously; while the higher performers are less likely to be able to find further optimizations without major paradigm shifts (in this case the phrase is actually appropriate).

We can surmise that so long as no physical limitation is put on intellectual output these same conditions will continue to apply. Everyone becomes more productive over time, but the less productive become more productive, faster than those who are most productive, because optimization of the most productive is more difficult; except when breakthroughs are made, when the cycle begins all over again.

The world of physical production is far more constrained.

Another example:

Although design and construction techniques have made housebuilding far faster than it was 20 years ago (force multipliers like the nailgun, construction adhesives, panelized construction, etc…), efficiencies in production technology will never allow a house to be constructed of 40% less wooden 2×4s.

We would need to shift to a new material, AND new designs to see that kind of efficiency; and even then the cost of adoption would be far higher until the replacement technologies became the dominant format within the distribution channel.

Quite simply, innovation without physical constraint proceeds geometrically (or faster), while innovation with physical constraint proceeds linearly or slower, with occasional leaps forward.

What you are speaking of when you say you think that we are unconstrained, or less constrained than Eric has stated, is that unconstrained intellectual development will trivialize constrained physical development.

In some areas, this has been the case… as Moores law has so dramatically proven for example; but in others it has not, and will not, without a massive paradigm shift.

The field of CPU manufacturing has been able to apply the force multiplier of unconstrained intellectual development; to compensate for the constrained physical development of fabrication processes, to the point of minimizing those constraints. Housebuilding on the other hand has not been able to make such compensations.

This is why CPUs are 1000 times faster and 90% cheaper than they were 25 years ago; but houses only cost 20% less per square foot to build (after accounting for inflation and normalizing for external factors on both sides).

The singularity will never happen, because the physical world is not amenable to being "solved" by infinite computing capacity.

Every major science and engineering problem in the world falls under these same parameters and constraints.

Now, if we can change fundamental science, to the point of matter reorganization (synthesis of arbitrary matter, from arbitrary matter, with reasonable energy inputs) then I’ll say the physical constraints won’t matter…

…but we don’t have any idea of how to do that, or if it is even possible.

…It’s a fundamental science problem, not an engineering problem. Every genius in the world could work on it 24 hours a day 7 days a week, and get no result, until and unless that fundamental breakthrough is made.

That is a physical constraint that cannot be overcome by intellectual effort; and there are many of them out there.

It’s why we don’t have fusion power, or direct conversion solar, or 1000mpg cars, or cars that run on water etc… etc… etc…

Once a fundamental science breakthrough IS made though, that’s when the unconstrained intellectual development of the engineers takes over, and the cycle begins again.

Oh and I should note before somebody gets stroppy that we’re not really sure if fusion power is a science problem or an engineering problem right now. As of right now we face three major problems with fusion:

1. We can’t figure out how to produce a stable, self sustaining reaction at a useful scale for industrial power production, and still control the reaction properly.

2. We can’t figure out how to get more energy out than we put in and still control the reaction properly (we know how to do it uncontrolled. It’s called a bomb).

3. We can’t figure out how to extract and use a useful percentage of the energy we would produce, if we could resolve the other two problems

Right now, we’re not really sure whether there is a fundamental problem we haven’t solved yet… or even discovered yet… or whether all of these issues can be resolved with engineering.
(On a side note... wouldn't this post title be a great name for a rock band?)