Safety towards and with gem based APM

From apm
Revision as of 14:20, 14 December 2023 by Apm (Talk | contribs) (removed the stub marker - Page seems pretty good now :))

Jump to: navigation, search

The advent of gemstone based APM (gem-gum-tec) will come with many dangers and opportunities.

Note that:

  • Discussion of runaway replication accidents is massively over-represented. And risk are typically overestimated.
  • Discussion of specialized weaponry is massively under-represented. And risk are typically underestimated.

Overregulation as a safety risk

Early misdirected overregulation can increase risk rather than mitigating it.
Government bending to public outcry from ill informed fear
without a proper understanding of the matter can backfire.

  • research can be driven underground (including under the cover of corporate proprietary trade secrecy)
  • access to early technology can limit access to new technology and foster social spread & two class society

There has been a somewhat related incident that
may have delayed the development towards gem based APM notably (without any benefits on safety).
It was not exactly overregulation by government-decreed law but
it was overzealous internal governance policy of a federal research program/agency.
See: The negative effects that public overexcitement can have and History

Importance of governance technology and insufficiency of mere technical measures

Safety in a world with AP manufacturing and technology
is a lot about about governance and related software technology rather than physical technology.
There are some physical technology aspects too, yes, but these alone can't suffice.

Specialized offensive weaponry

There is a serious dilemma here.
Discussing these things kind of overlaps with developing them or at least giving hints on how these could be developed.
As such the author is hesitant to invest effort in deeper research and more in depth explanations.
It would be highly speculative at this point anyway.

Just to give a superficial overview.
Especially high threat levels may be posed by the following:
(only once the technology is there - avoid premature overregulation)

  • insect to bird sized drones (bigger size makes them very fast in speed and accelerations despite viscous air or water)
  • microcomponent maintenance microbots applied maliciously (e.g. spying, sabotage, …)
  • weapons employing high amplitude sound shockwaves against humans (all too trivially easy to generate)
  • weapons employing EMP waves against unhardened machines
  • long range energy weapons like lasers and microwave
    especially in space (effectiveness of a nuke, precision of a scalpel)
  • complex multi unit multi scale systems

Purely defensive weaponry

Shields against high velocity impacts may fall in this category.
This involves environmental observation and active measures taken on impending impacts that are
both smarter and more effective than existing countermeasures like conter-explosions and hypervelocity whiffle shields.
Ultra high power densities of gem-gum systems may allows for directed counter of projectiles
so long the heat capacity can take the waste heat.

Somewhat related to defensive weaponry utility fog related systems might act
in an airbag like way to safe-keep against blunt lower speed impacts.
Also useful in a civil setting.

Proliferation – nuclear and chemical – actions of rouge individuals

The popular worries

A worry here is that gemstone based on-chip factories will make it easy to separate isotopes
concentrate fissile isotopes and allow any random rouge individual to build a nuke all by themselves in secrecy.

Several things here:

  • To gain sufficient fissile material one still needs to go through quite an amount of natural ore.
  • Nuclear plants and nuclear waste sites with higher per-concentrated material can be well guarded.
  • Such activity will elevate local radiation levels that can be detected.
  • A bad nuke will be more a dirty bomb than a nuke, bad enough, yes, but way below an actual nuke by output.

Lesser on the impact side but perhaps higher on the probability side of risk are

  • "mere" dirty bombs attacks and
  • chemical poison attacks.

A small easy to conceal device could produces high quantities of toxic gas in a short amount of time.
Then again gas attacks are possible even today as high pressue canisters can pack copious amounts of problematic gasses.
Such attacks unfortunately and regrettably have happened in the past. Like the Tokyo subway sarin gas attack in 1995 by the Aum Shinrikyo cult.
Supposedly the reason for why later some trash bins in subway stations were removed.

With gem-gum-tec advanced trash bins could immediately seal up and transport away to safe distance any rando stuff thrown in.
Doorways and existing train station pass-through checkpoints could also detect for radiation. This is difficult to hide.
(wiki-TODO: Compare lead shielded notable radiological material vs carrying some naturally radioactive bananas)

In summary it seems very very unlikely that every second city will be nuked every few yeas by some maniac.
There are many more factors beyond the isotope separation difficulty.
Organized groups may pose a bigger risk, but this is hardly new.

Possible regulatory measures

The following should be the exception rather than the rule.
If in any way possible a blacklist (all allowed but …) rather than a whitelist (nothing allowed but …).
The following stands in contrast to open source which always should be the default and the option considered first.

Make open sharing of critical highly dangerous information (information hazard) illegal to a high fine / prison time.
Yes, this can not give an absolute guarantee to prevent any and all leaks (via dark-web, lax clear-web, physical meetings).
but it can make the attainment of such information much much much more difficult.
Zero risk can never be attained but a very low risk can be striven for.

Also make the possession of such as hazardous classified information only legal
when the owner went through proper training receiving a license.

Misdirected efforts – Dystopial surveillance state

Making private communication that can not be surveilled illegal is not an option. Period.
Some 🤬 people are actually trying to achieve that by making end to end encryption illegal.
Worse even it could be made illegal to not carry an always turned on phone with you at all times.

Such things will do way more harm that the safety they could ever provide. Fact.
When problems are no longer properly communicated as negative feedback in the social control loop
because of self-censorship out of fear of livelyhood endangering backlash
then it's the beginning of the end of a working healthy society.

Yes there are some not removable tail risks involved when allowing for fully private communications.
But these tail risks we absolutely have to accept as facts of life. No way around it.

Runaway replication accidents – And intentional malicious release

Replicators as tool for malicious actors (unsuitable)

When intentionally optimizing for destructiveness for a terror attack
then one naturally ends up with what was discussed in the above section.
Specialized weaponry systems.
Much larger systems are much more effective and successful in self replication as:

  • they can move way faster
  • they can shield better against radiation like UV light
  • they can pack more capability
  • they can form disjunct multi unit collaborative systems where the units are task specialized

Replication runaway accidents. Can they happen? And if yes how likely?

Direct path:
During the bootstrapping the proto-assembler concept is outdated (since well before 2000).
Proposed is the development via early diamondoid nanosystem pixel (direct path).
Reaching systems capable of closed cycle self replication will likely cover significant chip area.
Several sides of the replication pentagon do not hold. Especially it's 2D on a chip so …
Risk of an runaway replication accident is not present.

Incremental path:
Free floating things like the foldamer printer may look like (and be confused with) proto-assemblers
but they are actually just primitive subsystems like a site activation positional stage.
Reaching systems capable of closed cycle self replication will be much bigger and possibly (quite likely even) not monolithic.
(As of 2023 we are still far away for this capability.)
Several sides of the replication pentagon do not hold.
Especially there's a dependence on special feedstock and
these foldamer assemblies are rather fragile needing water at proper temperature and such
Risk of an runaway replication accident is not present.

Synthetic biology (NOT on the APM development path and totally different far term target):
Direct usage of living cells or creation of cell like things (that by definition can self replicate) is not involved in this pathways.
The latter would be synthetic biology. A very very different by nature strictly separate branch of technology.
For that look into existing biotech risk managenment.
Risk of an runaway replication accident is present but the scope is well known from bio-hazards.

Advanced productive nanosystems:
Replicative nanobots are an outdated concept for that.
See: Molecular assemblers as advanced productive nanosystem - outdated
Still technology is advanced enough to make them by then just for the sake to show that they can be made.
They may be more micro than nano but still invisibly small like pollen.
Risk of an runaway replication accident is present but most likely well manageable by then.
Intentional malicious use for max damage leads to weapon systems instead.

Grey goo horror fable:
Horror scenarios of the whole biosphere being converted in mere hours are all 100% pure hype & fiction.
Even intentional say we wanted to convert the whole (unguarded) Venusian atmosphere
into safe solid silicon carbide and bind the oxygen to iron or something,
it would take centuries of highly directed effort. Alone from energetic limitations.
This would be specialized high performance replicative systems rather than replicating nanobots.

Safety measures against runaway replication

Generally safety measures can be applied at all of the six sides of the reproduction hexagon
via a combination of technical and governance mean.

Replication capability:
For a large stretch of non-replicative tech development and pre-scaling there won't even be that.
As of 2023 we are still well within this phase.

Building block availability:
This is one of the most popular of the six but also one of the least applicable.
It's pretty much a given of the technology level. Incremental path man often require reproduced parts.
Intentionally trying to add a dependency in advanced systems though is pointless as it can just be stripped out and
basically just amount to a software lock.

Energy supply:
For stuff on chip just pull the plug (not that it could escape the chip anyways see point "replicator mobility").
For foldamers systems in a vile it also should take very very difficult deliberate effort such that
replicators can extract energy form a natural environment like say sugar and oxygen from human blood.

Blueprint data mobility:
This is all about software control.
Physically storing the replicator blueprint within replicators just as
biological cells do it with DNA makes no sense from an efficiency standpoint even.

Replicator mobility:
This is a quite easy and effective one.
A system native on a chip surface just can't leave.
A system without active propulsion can't spread large distances in short time.

Sufficient adaptability:
As none of APM systems along the development pathway are based on living cells all these systems completely lack this capability.
Very advances systems may attain this capability but at larger scales.
This is more like artificial life then.
See gem-gum-goo, Artificial life, Zero sum situation

All of these may be more or less linkable with governance via software.
But better only once really needed. Avert premature lawmaking action.

Related

Retrieved from "http://apm.bplaced.net/w/index.php?title=Safety_towards_and_with_gem_based_APM&oldid=15878"