DUO Full Exploratory Diagnostic Report 6A: Preliminary;
Orgin: New Hope, Re-routed to Coruscate MILINT.
Testing Type: Strictly Non Destructive (SND)
Special Objectives
1: Determine Threat level of DUO.
2: Determine Utility level of DUO in service of the Republic Military
3: Determine Safety in use of DUO
4: Determine Orgin of DUO
5: Predict DUO Upkeep Cost/Benefit
6: Provide DUO Cost/Benefit analysis of partial or complete refurbishment of firmware or hardware
7: Provide DUO mission recommendation
Objective 1: Determine Threat level of DUO: report
When determining the threat a DUO brings to the republic, there are two portions worth considering. Functional Design, and the protocols and objectives of the droid itself. An HK unit provides an obvious threat, but a particularly cleverly programmed janitorial droid has and will again be responsible for the destruction of a major space vessel.
It appears the rough design process of this droid was “Make the computer core of a space station able to walk.” nearly 95% of its considerable mass is dedicated toward either processing large amounts of information rapidly, or powering or cooling the powerful computers. The limbs of the device are hardly even attached, and yet even those sport high performance heat sinks and pumps for coolant flow. In fact the only components that aren’t dedicated to the above are the “Forearm” modules, which sport micro fabricators and a stunning variety of tools, and the “foot” modules, which are both simply specially designed to allow the unit to walk on as many terrain types as possible.
These facts were initially quite alarming, but further analysis showed a number of factors that significantly mitigate the threat. Namely, Unlike any real attempt at replicating organic intelligence, the DUO is capable of a maximum of 4096 distinct protocol/process threads. This is significantly more than most droids, but far less than the millions or trillions normally seen in AI attempts. This also provides a rather low ultimate limit on the number of devices it can control at a time, which provides an indirect limit on the number of discrete units it can control or the total complexity. Additionally, Despite having a much lower capacity than any real AI attempt, it also is far too complex and has far too much available memory for this DUO to be cheaply manufactured. In fact, requiring a great variety of materials with unique characteristics, the likelihood of this model of DUO creation a replication crisis is incredibly negligible, as successfully blockading any one of a great variety of rare materials will block replication, and the loss of any core represents a value loss that would be devastating to any army that relied on them.
Once you get past its self-repair and security protocols, its motivational protocols basically amount to “Follow the laws and regulations of the Republic. Serve the Republic Military. Anticipate and prepare for future orders.”
If it weren’t for that last bit, this would practically be a classroom example of a military droid. Unfortunately it can only be so simple BECAUSE of that last bit. Recommend leave it intact, but monitor carefully. It will happily provide regular status updates on command, so do that, and annual random validations of those reports, and no threat exists.
Adddendum: Turns out the mission conflict with restraining bolts has already been predicted and designed around. The unit has a limited heuristic memory core design template available to it that can run while the bolt is installed. When removing the bolt you can choose to either integrate or discard the heuristic programming, and the modules can be mounted with or without the bolt attached, with a switch to set the module to read only or write only (One can of course simply install the module with no bolt for full access, or leave the module detached for no access.) In addition, these modules are stackable, and may be installed in any number, with a bolt sitting on the topmost one; though that can rapidly become quite unweildy and may even completely immobilize the droid.
Additionally: Despite coaxing, the droid appears to be conceptually incapable of anything properly resembling “slicing”. Bypassing protocols simply won’t occur to it. Either it’ll simply refuse, or if pressed it’ll do something destructive to complete its objective. Our final test was to retrieve the data from a locked server room in an adjacent station without damaging it in any way. The unit proceeded to jump across to the outside hull of that station and stay there, clinging to the outside of the station while it fabricated a small remote controlled droid to disassemble its way into the station, neatly remove the requested server’s hard drive, and physically carry the objective back to the assigned handler. This was after being loaded with remote access protocols, which went almost entirely unused. These two facts combine to dramatically further reduce the risk this unit represents.
Objective 1 Report Summary: Physical threat negligible, Intelligence threat mild, Logistical threat: unlimited Treat like a laser drill. Technically it can cut through anything, but it’s easily blocked, destroyed, and thwarted. Use it like a tool and it’ll serve you forever. Do not permit enemy to fall into enemy hands.
Objective 2: Determine Utility level of DUO in service of the Republic Military: Report
Preliminary Report: Unit has demonstrated a feature set that allows it to assist to a majority extent, if not perform entirely independantly on command, nearly any task that might be set to it on a repair yard, hover field, or droid depot. Equipped with a number of tools considered far too expensive to mount on a motile droid unit, and a comprehensive (and now upgraded) engineering protocol set, This unit can be used to complete tasks that would normally require docking at a major facility to manage. Better yet, the manufacturer claims that if properly supplied, this unit can be deployed to an iron rich system and build a functional starbase in only 4x the time of similar constructions bids, while requiring only a single resupply at approximately the halfway point as opposed to the normal constant supply stream required by any manned facility. It would be a disservice to the Republic to let it go to waste, though of course no final recommendations or orders will be issued until such time as this report has been completed.
Following In Progress/Notes
3/3
||REPMILINT2: Sort of feels like cheating, to get the answers like that, doesn’t it?
REPMILINT1: What? How so?
REPMILINT2: Like looking at the answer at the back of a crossword isn’t it? We do all the work to get started and the specification sheet just falls into our laps.
REPMILINT1: Well if you want you can do the rest of the process
REPMILINT2: No, it won’t be the same, I already know all the answers
REPMILINT1: So wait, you don’t look at the answers at the back of a crossword?
1/3
||REPMILINT1: High
REPMILINT2: How High?
REPMILINT1: Very High.
REPMILINT2: We can’t just say very high.
REPMILINT2: [REPMILINT1 NAME]?
REPMILINT1: It’s kinda scary
REPMILINT2: It is.
REPMILINT1: …High||
Objective 2 Preliminary report Summary: More than sufficient to justify applicable risks, if assigned in close co-operation with any senior technician.
Objective 3: Determine Safety in use of DUO
Consideration of the safety in use or republic equipment covers the safety involved not only in normal operation, but also operation when misused, or while damaged.
In both fields the DUO performs excellently. When properly assembled, serviced, and repaired, all potentially hazardous materials are in well designed containers contained within armored compartments, and unless specifically ordered not to, unit will perform repairs and servicing on itself given the merest scraps of idle time. Overall, in this matter, this droid performs better with the functions of both a battle droid and a medical droid than a great many units we have assessed that only had one of those capabilities.
Notes
||REPMILINT2: No radioactives, a number of caustic chemicals in tanks in the toolsets, absurdly powerful laser that can be configured on the fly to function as an okay blaster pistol, Suggest when salvaing destroyed or damaged units leave the forearms behind unless a full HAZMAT team is available, otherwise once the code’s checked this should be fine on the field. Useful too, wish I had one of these behind me when I got blown up
REPMILINT1: Nothing suprising in the software either. If this unit becomes a danger, it’s going to be because of what it built, not what it’s made of or doing||
Objective 3 report Summary: Safe
Objective 4: Determine Orgin of DUO
Preliminary Report
We have fully verified the droid’s orgin as the invention of the brains behind Bonshyyyr Incoroperated, based out of what was reported to be a decomissioned starship on Umbara for this research project. We have located and have available for command review all project documents and a historical timeline. The only thing holding us from marking this objective complete is a three month timeline gap between departing the prototype lab and the security purge event, which still has a great many open questions. Unfortunately, it appears that particular investigation will take some time and co-operation with Planetary Civilian CI which, considering one of the most critical events in question took place on Hutlar, provides a number of issues. Please refer to report RMPI: Missing Property: CRF 90-90 for further details as investigation progresses.
Prior Preliminary Report:
We have confirmed this came from one of our own R&D projects. Looks like a division of these units was sent out to New Hope for practical testing, but there was some sort of logistical failure and not only did the unit vanish without being noticed, the orders never found their way to New Hope for receiving. We have verified payment, We’ve received a number of progress reports, and we have the tracking number for the delivery, which reads as having been held up at Hutlar, I might add. We are attempting to locate the people directly involved in R&D, A contracted five man team known as Bonshyyyr Incorporated, but we have not yet received any response. Please be aware: You may or may not receive or locate further hardware. This stage of testing requires a company’s worth of material, but the mass listed on the delivery package roughly matches the weight of the DUO in question, plus or minus a couple of its modules. We should have those security charge reset protocols for you soon.
Following In Progress/Notes: Page 6/12
||REPMILINT2: Here’s the tracking number for the shipment.
REPMILINT1: What shipment?
REPMILINT2: The practical testing material delivery tracking ticket on our logistical registry.
REPMILINT1: What? They haven’t even actually gotten the approval notice yet, it was sent not even twelve hours ago.
REPMILINT2: Generated it days ago. I suppose they already had the product and jumped the gun
REPMILINT1: Practical testing gives two years for manufacturing before even asking for delivery
REPMILINT2: You know what I think? I think they didn’t know that.||
Following In Progress/Notes: Page 2/12
||REPMILINT2: I was riiiiight!
REPMILINT1: No you weren’t.
REPMILINT2: Then what’s this then?
REPMILINT1: That is pause, then a deep sigh That is a memory stick.
REPMILINT2: And what’s on it?
REPMILINT1: It’s the unit’s serial number isn’t it?
REPMILINT2: It is. And you know where I found it?
REPMILINT1: Does it matter?
REPMILINT2: It does to me. It was plugged into auxiliary storage 3. Which means…
REPMILINT1: It means you’re right. Is that what you wanted to hear?||
Objective 4 Report Summary: Republic R&D programs
Objective 5: Predict DUO Upkeep Cost/Benefit
The redesignated CRF 90-90 droid is incredibly expensive to manufacture, but once the first is completed, construction of any additional becomes mostly a question of timeframe. Additionally, the unit is capable of maintaining itself as well as other models of its type to some rather exacting specifications.
The only real upkeep for this unit, assuming it is given sufficient time to perform self-maintenance as it desires, is a monocrystalline diamond that forms the core of the primary plasma cutter on its left hand, and, in addition to having one spare already embedded in its chassis, they only will need replacing in case of some incredibly specific damage or an estimated fifty years of heavy use.
Thus, while loss of this unit would represent an irreplacible loss, actual upkeep costs are negligible. A second, or better yet, a squadron of five of these units would shift the balance of this calculation further in the droid’s favor, due to their ability to salvage and fabricate components for each other.
Its benefits are extensive and already detailed in the other portions of this report, but we do feel that special warning should be issued that this unit being left to operate unmonitored for extended periods of time continues to represent an unacceptable risk. These units should be set to provide constant remotely-accessable logs, which should be error checked annually, and a twenty year refit/replace cycle is recommended.
Coincidentally enough, its maintenance schedule takes a fraction of its own available work schedule that is within the same order of magnitude as our own R&R schedule in an extended warefare setting. It is for this reason that we are also recommending that local unit commanders simply include the droid unit on their relief schedule. This will give it time to maintain itself, provide dumping of all the logging, and still leave it time to wander amongst the crew making small goods to improve morale.
Following In Progress/Notes
Page 4 of 5:
REPMILINT2: So with one exception, everything it uses it can fabricate itself entirely from raw materials?
REPMILINT1: So long as it can find materials to make a modern forge.
REPMILINT2: So long as that, yes. So what actual upkeep cost is there?
REPMILINT2: Other than time, I means
REPMILINT1: It can’t make another diamond until after it’s made pretty much everything else, and eventually it’ll wear out.
REPMILINT2: The Monocrystalline diamond will eventually wear out?
REPMILINT1: Well, yes.
REPMILINT2: And that’s the only consumable
REPMILINT1: Assuming you give it enough time and raw resources to work with.
REPMILINT2: Or the tools to gather the resources.
REPMILINT1: Or… Oh yeah. I’m going to have to rewrite this again now, aren’t I?
Page 1 of 5
||REPMILINT1: This thing can self replicate, but it’s slow, it can think faster than us and can remember everything, but it’s stupid and not very creative. Mostly it’s incredibly expensive, which both limits its risks and upkeep costs
REPMILINT2: Actually, Look at these tools.
REPMILING1: Yes, it can make and fix… nearly anything, I suppose. Wow, is that a 1 micron monocrystal diamond? Who could possibly thinks those should be able to just walk around?
REPMILINT2: Someone who thinks this droid should be able to fix its own tools.
REPMILING1: That… That’s why the supercomputer has cargo mounting points, isn’t it?
REPMILINT2: Like you said, they’re expensive.
REPMILINT1: You check on how well that stuff stores and I’ll check how much it costs.
Objective 5: report Summary: Extreme Efficiency for upkeep required: Use where possible.
Objective 6: Provide DUO Cost/Benefit analysis of partial or complete refurbishment of firmware or hardware
This one’s a nonstarter immediately out of the gate. It uses parallel memory for everything. Too complex for droid memory and too simply for server computer systems, everything has to be converted, purpose wrote, or heuristically adapted for it. There’s a reason no one uses base twelve memory values. As far as the hardware is concerned, this thing represents enough rare materials to make a dozen astromechs, and yet it’s complex enough that recycling would be an incredible excercize in frustration. About the only thing it’d be good for is to mount it into something else, but good luck maintaining it without another of its model or a fully staffed engineering team.
Objective 6: report Summary: Either leave it alone or mount it to a Consular class to reduce the crew requirements by around half for double the cost.
7: Provide DUO mission recommendation:
Recommend installing OEM Control packages attached to this report into core.
Recommend patching core with Module labeled “Security Purge Recovery”, after OEM Control Package installation, if utilized.
Recommend cloning “Security Purge Recovery” and distribute to all officers commanding units that might engage with a CRF 90 model droid after a security purge. (Currently: Fort Hope Command structure)
Recommend assigning a logistical support role in partnership with local depot commander. Details to be determined locally
Recommend providing access to full stock of droid creation materials to facilitate self-replication.
Recommend sending first, second, and third “Successful” attempts at self-replication to REPMILINT facilities for further researching, development assessments, and destructive diagnostics.
Recommend at least three model CRF 90-90 droids remain locally available if and when fabrication facilities for creation exist, one primary, and a final for unexpected failure.
Recommend training all officers and unit commanders that are expected to engage in close operation with CRF 90 units on Security Purge and Combat Security Purge usage and operation. Training can be provided by CRF 90-90 without concern.
Recommend constant logging via wireless stream. Logging should be error checked using physical methods weekly for at least 3 months, followed by annually for droid lifetime.
Recommend assigned handler retrained on most up-to-date available Heurestic, Droid, and Networking, safety and security proceedures on a bi-annual basis if not already.
Recommend assessment and intial promotion fast track for specific purposes of command of other droid units, once sufficient usagage to satisfactorily prove reliability exists.
Recommend Military Restraining Bolt be used during motile operations, and physical attachment to base be used during design applications. Both may be forgone in case of urgent combat operations, so long as data dump has been manually checked for heuristic drift or core protocol corruption/modification recently.
Prior Preliminary mission recommendations:
We’ve completed the physical analysis and are returning the unit now.
Recommendation 1: Standing orders to run self-maintenance quietly in a closet to itself with the bolt installed until the rest of this report is complete. Wireless and linguistic modules has been removed temporarily to permit for assistance as technical droid in case of manpower shortages however.
Recommendation 2: Briefly retrain all officers on signs of sentience development, rampancy, and protocol drift.
Recommendation 3: Train all officers on the usage of the unit’s integrated “Security Purge” feature. Train all high officers on the existance and appropriate use of “Combat Security Purge” feature, which will result in DUO exploding after no less than ten and no greater than fifteen seconds. Yield is currently uncertain but explosion will likely be sufficient to damage or disable light vehicles, and will most certainly destroy the DUO.
Other: Attached holocom message may be played safetly, as desired.
Objective 7: report Summary: Logistical support role. See full report for details
Report Notes/Comments/Remarks
Associated Reports: RMPI: Missing Property: CRF 90-90
REPMIL COMMAND: Unit installed
REPMILINT1: If we can get just a couple dozen of this unit, it has the potential to completely re-write space-based logistical doctrine. It could also drown us in metal if given sufficient time to build up. At this point, it is my opinion that we must allow it to self-replicate as rapidly as possible to get ahead of an exponential curve. Once we have a dozen or so we can re-assess the risks and proper use, but as of now, the exponential nature of the algorithms make it hard to determine if it will even actually be able to increase its own count, or if wear-based or unexpected failure will terminate and its line before a population really gets started. And yes, I am well aware of how terrifying it is to use the words “Population” and “Generation” in this context
REPMILINT2: As impressive as it is at a glance, it neither provides any truly new capabilities or even particularly imaginative engineering. Everything it can do can be done either faster or more effectively by other available droids or processes, it’s merely the fact that this unit can do ALL of it that makes it so complex and expensive. Personally, I expect it to fail catastrophically before it truly manages to self-replicate, but I almost hope to be proven wrong.
REPMILINT1: It’s not one of the objectives, but it makes for an amusing anectdote. The programmer for this was paranoid, and implented a hardware level magnetic pulse charge, alongside a block of explosive, that can be toggled independantly of eachother, for if it detects unauthorized tampering. One of the triggers for the memory charge is if the serial number doesn’t match an encryption protocol mounted to it. It was issued valid republic communications codes, Experimental for testing so they wouldn’t have even worked for anything besides IFF, but valid encryption. But somebody, almost certainly one of the civilian researchers had removed it for some reason, and since, apparently, all the memory interfaces on it are standardized to an extent, when he plugged it in, he plugged it into an auxiliary storage unit. This would’ve been fine, had the serial port been empty, but it was not empty. Instead, what was in it what appears to be a HoloComm message for the unit commander. We’ve placed it into an auxiliary memory bay and planted the restraining bolt where the serial number goes, for now. Also, We’ve found the precise instant the problems with this unit started. See attached network reference.