The horror fable of the grey goo.
Picture: My Little Pony - The Smooze

Knowns:

SciFi shocker goo-ball earth: impossible
accident: impossible in the development process of advanced productive nanosystems
mild forms: probable after the development of advanced productive nanosystems (See: Reproduction hexagon)

Definition

The widespread knoweledge about the hypothetical gray goo scenario was born from the gonzo image that some "nanobot" like form of productive nanosystem fails to stop replication and consumes the whole surface of the earth leaving only gray goo behind. This superficial idea self replicated in the public mind like a virus. It can thus be called a successful meme. It did quite some unjustified reputation damage (see: History).

Current expert knowledge is:

An accident like this is with a probability bordering on certainty imposible in the early development stages.
Deliberate development will start to become possible and rise in probability once technology level III is reached and will reach near certainty on further development. It will bring limited forms of this problem.
If official development will not be prohibited, there will be means for easy containment in place.

One must not forget that there are other dangers that are equally or more dangerous than uncontrolled replication. If it is so dangerous why APM then? Simply because there's a huge list of opportunities and products potentially improving our and restoring natures world.

Factors for large scale unbounded replication

The reproduction hexagon as an analog to the combustion triangle

Requirements

There are several factors which all must be true at the same time for an accelerating chain reaction to occur. In the fire department there one has the combustion triangle. With replicators one can do a similar thing and create an replication hexagon (since there are around six requirements which must hold simultaneously.

Replicativity

In early productive nanosystems self replicapility could be avoided by building many active components in parallel onto a chips surface.
Whether it is easier to create a T.Level I Nanofactory with replicativity or parallelism remains to be seen.

In T.Level III productive nanosystems of the classic (deprecated) assembler type need to be replicative to produce macroscopic amounts of material.
Advanced nanofactories (in form of solid bricks) will almost certainly have self replication capability (think of 3D printing a printer chip).

Medical tools (in form of dispersed nanomachines) can be nonreplicative products this is the best security measure one can get. It may be reasonable to in general prohibit dispersion of nanosystems with replicative capability in liquids or gasses, but at some arbitrary size level a line has to be drawn.

Resources

In technology level I it is quite hard to produce the building blocks. The simple assembler mechanisms (at this stage one can think of them as linkages) can only put those whole preproduced blocks together.
In technology level II It's not yet clear which resources will be used.
In technology level III the not natural occurring ethine (= welding gas = HCCH) among other unnatural resource materials will be used.

Disassembly of nonstiff disordered natural substances is way harder then assembly of carefully chosen synthetic ones. In fact in many cases it's as good as impossible at room temperature. The humic substances (TODO: add picture) for example are hopelessly to complex to use them as resource material. Certain minerals may be directly disassemblable with substantial effort in the not so near future. The carbon dioxide in the air constitutes a vast reservoir of easily attainable building material. This source could pose a threat.
To investigate: minimal atmosphere-floating solar replicators Warning! you are moving into more speculative areas.

Energy

Productive nanosystems will likely get their energy from the chips surface - a good security measure.

In some cases energy can be drawn from the resource molecules. Dissolved sugar would be such a case, but it is not a good resource since it has way too much hydrogen. Its low stiffness may be a problem too.
In others not (carbon dioxide)

Mobiliy

Productive nanosystems will always have their components firmly linked together in machine phase. Only macroscopic blocks will separate.
[todo: describe: policy exceptions limits] mobility prevention guideline

Exponential assembly puts an absolute insurmountable limit to spacial spread of replicated structure.

Mutations

AP systems do not evolve by themselves. A mutation of a productive nanosystem to a new ecosystem is as likely as that firefox mutates to firegodzilla and infiltrates new operating systems. Due to the very different architecture of living organisms and APM systems radiation has a completely different effect on them.

Productive nanosystems may loose efficiency or function. Logical errors (which still are functional) are extremely unlikely.
Biological systems also loose efficiency and function but have a non neglectable chance to suffer still functional logical errors aka mutation instead

A good description of part of the situation: alive-ness of computer viruses
Also: can computer viruses evolve
Or: "Nanofuture" chapter 8 self replication by J. Storr Hall

Replication data

Is likely to be stored semi locally (millimeter scale) since it's convenient and disaster proof.

Antagonistic phenomena

Radiation damage as limiting factor

Especially in the atmosphere (with plenty of solar power and carbon dioxide raw material present) UV radiation that is short enough in wavelength to split even strong chemical bonds constitutes an additional limiting factor.

In the hydrosphere radiation is quickly filtered out by water when going deeper. But light is absorbed too, so a power source is missing.

Active detections and countermeasures

TODO ...

Macroscale "goo"

Main article: "gem-gum goo"

Radiation shielding, efficient long range movement for the gathering of energy and processing of more diverse resources all that can be archived much easier by much bigger replicative units (easily visible by the human eye). This is exactly the strategy the evolution of biological life has followed. But this would not be gray goo in the usual sense anymore.

This would be a scenario with much more sophisticated abiotic macro-replicators. It would require an extremely high development effort. If such replicators are omnivoric and adaptive enough, they may even deserve to be called an (invasive) abiotic artificial life form.

Accidental ingestions of these big replicators is less likely (only from decaying broken down units).
Catching few big replicators is obviously easier than many very small ones (except they feature self defense)
A likely development path may be via artificial pets for company mimicking real ones as closely as possible (ultra advanced "Aibo" robo-dog) then maliciously altered into replicators.

Big things have a big design space. So on the long run this scenario may lead to kind of a highly diverse parallel mechanosphere ecology.

Since our biosphere ecology would be defenseless against such a mechanosphere ecology (biology never evolved to eat rock the same way it eats hydrocarbons and likely never will) there will be a need to design systems to safe guard biosphere ecology. These systems may be part of the mechanosphere ecology itself (in sake of long term stability) or may be run in a more centralized control (in sake of keeping more control). Given the possibility of high intellingence for big replicating units the lines blur.

Biological analogies in earths history

On the long term in natures history unbounded self replication never lead to ultimate destruction by mono-cultures. Instead speciation lead to beautiful complexity. In how far can this be applied to artificial self replication? See: gem gum goo

In the past

Major empty (biological) ecological niches naturally where present in the early history of earth when earth was just "empty". Cyanobacteria came upon a carbon dioxide rich atmosphere above the seas and pretty much no competitors.

This lead to "the great oxygenation event". The accumulation of their excretion product oxygen in the atmosphere

back then oxygen which was toxic for many lifeforms.
the oxygen "burned" the methane in the atmosphere to carbon dioxide a much weaker greenhouse gas inducing the longest ice age in earths history (the Huronian glaciation)

A mass extinction followed.

Later in earths history it usually took major extinction events (not induced by life itself e.g. asteroid impacts) to clear some space (side-note: indicating life already ran into a "permanent" malthusian trap). The species most successful in filling in the new gap are prone to trample over even more species in their rampant growth. Species that otherwise would have survived the extinction event itself got (and get) wiped out. (On a much smaller scales a similar effect can be observed after a fire. Especially where fires are rare.)

In the carbonifeous period scale trees (strange looking compared to todays trees) for the first time produced lignin and that in large amounts. Its accumulation may well have had a extinction effect on many species. The first species able to eat and digest lignin (ending the carbonifeous) again might have had a period of explosive growth into the new niche with detrimental effects for other organisms.

Today - humans civisiastion as bio-goo or seeding bio-goo of feeding bio-goo

With our excessive agriculture great amounts of fertilizer are swept down the rivers into the sea this has induced unnaturally massive algae blooms leading to depletion of solvated oxygen in the water and consequently the emergence of horrifying dead zones reaching over vast expanses.

With our unnatural means of transport we often bring invasive species to new habitats (e.g. rats to almost everywhere, rabbits to australia, ...).

With our burning of oil and releasing as carbon dioxide oceans become more warm and acidic. All the organisms that use calcium carbonate for their structure (shells) dissolve and get wiped out. On land the effects are not as clear but change at this rapid rate will lead to an extinction / diversity loss that is much faster than evolution can compensate with buildup of new diversity.

(Side-note: "Our" crude oil stems from the aforementioned scaletrees. It's all the stuff that got buried and sealed away before lignin eaters evolved into existence and ate up all the stuff that came later after the carbonifeous period.) The most important resource ....

Future

The things about to come are fundamentally different in the way that some parts of gemstone based machinery (nano & macro) are fundamentally incompatible with the biosphere. Some parts behaves more like undegradable rocks. Evolution, while having plenty of time to try, never succeeded in using the lithosphere to the same degree it uses the hydrocarbon centered biosphere.

general pattern

Problems explosive growth can cause can be grouped in three categories:

depletion of needed resources
accumulation of troublesome waste
direct attack (pandemics)

Figurative analogies with water:

Dam break: a deeper empty reservoir symbolizing a big ecological niche that is empty for "the water" but filled with other value that gets devastated by "the water".
Sand dam break: a small trifle becoming increasingly faster until its a raging stream (exponential growth up to some point)
Water hammer: filling an empty void is fine but when the limits are reached some kind of abstract "inertia" can lead to a destructively overshoot

Examples for replicators from deliberate malicious intent

Microscale airborn carbon dioxide utilizing replicators

A form of those is called aerovores (name coined by J. Storrs Hall). They are solar powered air "breathing / eating (vore)" replicators. Note the (mis)use of biological analogy here.

much desin effort needed: minimal nanofactory grains will probably be too big more assember like system need to be developed
much desin effort needed: restriction to atmospheric elements H C N O and traces of others
brown movement allows small particles to fall very slowly slight (solar) heating lets them rise
when they emerge countermeasures will be in place (e.g. skysweepers - J. Storrs Hall) that is effects of malicious AP systems must not be considered in context of todays technology. (unilateral disarmment through suppression of research imaginable)
there is a serious health hazard of breathing even a few of them in
they can be rained out and washed into the ground
effect of radiation to time of operationality (TODO: find list and extend research about radiation effect on microscale self replicating devices)
self delimiting process - dangerous CO2 depletion
lack of adaptability - brittleness of artificial systems
spill - misconceptions

Other

Warning! you are moving into more speculative areas.

Atmospheric units that use CO2 as building material and sunlight as energy source potentially endangering plant life. In the worst case those thing could silently create huge amounts of Hydrogen Cyanide or NOx and release it in one blow.
Some form of green goo medical tools with replicative capabilities that uses sugar and other common bio-molecules as resource. Medical tools will by definition be in form of seperated particles suspended in fluid phase and thus much more mobile. There are no known investigations to break up sugar for this process.
When there is a whole global network of an AP system the hole system could be infected by a computer virus creating all sorts of strange effects from useful or funny over weird to really bad stuff. (Related: techno plants)
loose nanomachines (from advances in technology level III) that combine the capabilities of medical devices with very compact molecular assembler like self replication capability and can use e.g. sugar or ATP as carbon and energy resources.

Similar phenomena

Microcomponent pipeline breach

In the case of the existance of a global microcomponent redistribution system there could be something happening that amounts to a software based pipe breach. The cause could be a software bug on top of very flawed software design or some malicious computer virus. Any non mobile stuff thats expulsed may eventually sooner or later clog the outlet. Specifically programmed utility fog though could flood a wide area. As long as the foglets don't leave one connected machine phase they may be easily retractable when the software problem is resolved. Pushout force safety limits should prevent mountain high pileup of material till it crushes under it's own weight.

Disassembly to microcomponents attack

Warning! you are moving into more speculative areas.

A (not desirable) system combining the capabilities of microcomponent maintainance units with utility fog could in thin layers flow around objects that are constituted from microcomponents and start disassembling it - intended or not. To defend against this attack physically one can use hierarchical locking for the outer surface of parts which converges to just a few combination lock stones.

External links

Wikipedia: Great Oxygenation Event

Videos

PBS eons: "That Time Oxygen Killed Almost Everything" [1]
Purple goo / Purple earth hypothesis: PBS eons: "When The Earth Was Purple" [2]

Grey goo horror fable

Contents