Tech Tuesday 01/02/2018 New Years Edition

Oscar Gordon

A Navy Turbine Tech who learned to spin wrenches on old cars, Oscar has since been trained as an Engineer & Software Developer & now writes tools for other engineers. When not in his shop or at work, he can be found spending time with his family, gardening, hiking, kayaking, gaming, or whatever strikes his fancy & fits in the budget.

Related Post Roulette

55 Responses

  1. Pinky says:

    Comp2 – The Caliburger story could qualify as a supervillian origin story, along the lines of a Brainiac or Ultron.Report

    • Oscar Gordon in reply to Pinky says:

      Calitron is rampaging through downtown L.A. destroying all it sees, and occasionally stopping to scream things like, “Why don’t you want avocado mayo on your chicken sandwich, James?! Everyone in California loves Avocado! DIE!”, at fleeing pedestrians, before vaporizing them with jets of boiling fry oil.Report

      • Chip Daniels in reply to Oscar Gordon says:

        In all seriousness, given what we know about how AI inevitably takes on the biases of its creators, how the combination of facial recognition software and AI handles interaction with different ethnic groups once it becomes ubiquitous, warrants a fair amount of caution.Report

        • Charlie Stross has a new and depressing piece up at his blog about how bad things could be with AI only a bit faster and cheaper than our current state of technology.Report

          • Chip Daniels in reply to Michael Cain says:

            Thanks for the link; It is disconcerting.

            I binge watched the first couple seasons of Black Mirror but had to stop since it was becoming seriously depressing.

            The consistent villain in it was not the technology, but the weakness and cruelty inherent in humans themselves, how technology just became a force multiplier of our worst aspects.Report

        • Oscar Gordon in reply to Chip Daniels says:

          There is a social aspect as well. If a person can interact with a kiosk rather than a person, they can avoid developing a lot of social niceties and tolerances for people not like them. So no longer does the racist have to deal with the black kid behind the counter and keep his bigotry in check, when he can deal with the always pleasant kiosk AI.Report

      • Pinky in reply to Oscar Gordon says:

        I’m thinking more about the day that your burger joint’s AI starts complaining to your insurance company’s AI about your newfound love of cheddar topping. How can an insurance company justify not gathering that kind of information? And if we can sue cigarette companies for our cancer, why can’t we sue fast food chains for our weight gain?

        “I want large fries with that.”
        “I’m sorry, Dave, I’m afraid I can’t do that.”Report

        • Chip Daniels in reply to Pinky says:

          @pinky @michael-cain
          We ourselves struggle with concepts of fairness and justice, yet we are offloading more and more human management tasks to algorithms and machine learning.

          Who gets priority for a job, an apartment, a space in a school; Who is at risk for crime, poor health, child abuse, bad credit; Who gets a promotion or dismissal; Which community should get an upgrade to its infrastructure and associated jobs.

          We can’t possibly tell the algorithm to produce a just outcome because we ourselves don’t even know what it looks like.Report

          • Oscar Gordon in reply to Chip Daniels says:

            If just outcomes cost nothing except the effort of making the right decision, this wouldn’t be such an issue.Report

          • Nor, in general, are we going to know precisely what the algorithms are learning/training themselves to do.

            Someone brought up the AlphaZero game-playing software (self-teaches all of Go, chess, and shogi playing against itself) the other day. It’s better than human play has been described as “chess from another dimension” and “Go from the future” by some of the best human players because the program wins doing things that make no sense to people.

            One of the things Stross describes are Facebook experiments with identifying people even though they don’t have a Facebook account from indirect things like picture tags, textual references, etc. Should such a person get a Facebook account, the software will already know considerable about them.

            Goal-seeking software and big data are a scary combination.Report

            • Some months back LinkedIn spent a few weeks asking me if I knew someone who was our own Will Truman under his real name. From time to time I wonder if any of the people they offer to hook me up with is Burt Likko’s meat-space self.Report

              • Chip Daniels in reply to Michael Cain says:

                What’s startling is when Facebook shows me a picture of my ex-wife as “People You Might Know”.Report

              • Troublesome Frog in reply to Chip Daniels says:

                Pinterest just started offering my wife Vietnamese translations for some reason despite the fact that she has never given Pinterest any hint that she’s a native speaker. We’re still trying to figure out how that information leaked onto Pinterest’s servers.Report

          • Pinky in reply to Chip Daniels says:

            I’ve argued that we’re already using algorithms for such things; we just call them laws and regulations.Report

            • Kazzy in reply to Pinky says:

              I had a long discussion with a colleague recently about the role of technology in the classroom. My position was that we can’t really talk about “technology” without defining what “technology” is. Right now, when we’re talking about technology with younger children, we really mean screens and little else. So if we want to talk about screens, let’s talk about screens. But in reality, especially for young children, “technology” is damn near any tool in existence, since almost all of them are new or novel. At which point I argue that technology should be used for two primary purposes: to make the impossible possible or to make the already possible better, easier, more effective, more efficient.

              Then everyone wants to talk about screens and apps again.Report

          • Oscar Gordon in reply to Chip Daniels says:

            Speaking of big data and all that, occasionally it can be used for good.

            I meant to include this in the OP but forgot about it.Report

          • j r in reply to Chip Daniels says:

            We can’t possibly tell the algorithm to produce a just outcome because we ourselves don’t even know what it looks like.

            I think that’s the whole point of having machines do some of these things.Report

        • Oscar Gordon in reply to Pinky says:

          “I want large fries with that.”
          “YOU WANT FRIES WITH THAT?! DO YA, TUBBY!? DROP AND GIVE ME 20! THEN MAYBE I”LL LET YOU ORDER FRIES!”

          (For the Caliburgers on USMC bases, Gunny Ermey supplies the voice)Report

        • North in reply to Pinky says:

          Somehow I think your burger joint would find a lot of reasons not to let that happen. Mostly green crinkly keep our jobs and our existence reasons.Report

  2. Pinky says:

    Arch1 & 2 – These are the two extremes in terms of windstorms.Report

  3. Michael Cain says:

    Aero1: How much of this is an admission that carriers are going to have to stand off at much greater distances in order to remain safe in the future?Report

    • Oscar Gordon in reply to Michael Cain says:

      @michael-cain

      I don’t know what you are talking about. No one has any clue what you are talking about. Carriers are the preeminent weapons platform in the US arsenal, second only to nuclear weapons.Report

      • Maribou in reply to Oscar Gordon says:

        @oscar-gordon

        “No one has any clue what you are talking about.”

        That was a bit harsh/unnecessary.Report

        • Michael Cain in reply to Maribou says:

          I took it as sarcasm and was not offended.

          Anyone who looks at military tech has at least thought about the question, “What kinds of anti-ship tech could keep carriers at a distance outside the round-trip range of their planes?” Certainly the US Navy research program thinks about it: aerial refueling, laser weapons, and rail guns (as well as the greatly augmented on-board electrical generating capacity to make all those things feasible) are all responses to that kind of threat.Report

        • Oscar Gordon in reply to Maribou says:

          @maribou

          Sorry, forgot the sarcasm tag.

          As I’ve mentioned before, in the Navy, carriers are referred to as LSTs, Long Slow Targets. Their primary weapon(s) are the aircraft they launch and recover, but their defensive ability largely rests upon the Aegis systems in the cruisers and destroyers that act as their screening elements. So it isn’t the carrier itself that is the actual weapon system, but the fleet that is centered on the carrier.

          Kolohe might know better than I, but I remember rumors that both Russia and China have been developing cruise missiles that are fast and smart enough to work their way through an Aegis screen and hit a carrier (something the Navy is not happy to admit to, hence my sarcasm), which would give the Navy a strong incentive to be able to extend the range of their air wing without having to rely on the USAF or Navy tankers that are limited to land based airfields (tankers that are notoriously un-stealthy and slow and easy pickings for a hostile air force).Report

          • Kolohe in reply to Oscar Gordon says:

            this was on the tweets today, as was this, but it’s more about land attack capability than A2AD at sea.

            On the Chinese side, the carrier-killer is supposedly the latest version of the D-21 ballistic missile, but my knowledge on both sides of the equation is woefully out of date.Report

          • Morat20 in reply to Oscar Gordon says:

            Carriers died the first time a drone fired on a target. They just don’t know it yet.

            Hell, the modern Navy is a dead beast walking, as is the Air Force.

            Their combat roles can be better filled by swarms of much cheaper, semi-autonomous drones operating under the control of coordinators hundreds or thousands of miles away. The Navy, commanded by “captains” sitting in a chair on dry land, or the Air Force flown by “pilots” in front of monitors in Nevada.Report

            • Oscar Gordon in reply to Morat20 says:

              I remember the discussions when it was obvious that drone technology was not going to be the sole domain of the US for very long. The Navy knows, the USAF I’m not so sure about (their egos keep getting in the way).Report

              • Morat20 in reply to Oscar Gordon says:

                The current big roadblocks I can see are AI limits (because, you know, jamming is a thing) and getting rid of the need for GPS.

                In a real stand-up fight with someone roughly on our level, communication will often be spotty to non-existent, and the GPS network is going to be hammered or absolutely gone.

                But even now, other than the initial grab for air superiority (ie, when all the jamming and GPS busting will be at it’s peak), you could replace the rest of the AF with drones.

                Just have them loitering around combat areas, beaming back intel and ready to be directed to hot spots as needed.

                As for the Navy — drone carriers alone would be a pretty big win. More projected power, smaller and cheaper ship. Life support ain’t cheap. Toilets and food and clean water and bunks all that other messy crap.Report

              • greginak in reply to Morat20 says:

                Can drones be protected from EMP?Report

              • Morat20 in reply to greginak says:

                Well, let’s put it this way — they’d not have it any worse than what the air force is flying now. Or what the Navy is sailing.Report

              • Oscar Gordon in reply to greginak says:

                EMP hardening is a thing, but it adds weight. Modern military aircraft can afford to carry that weight, but drones might not be able to afford it as easily. Of course, nobody has a directed energy EMP weapon yet (that we know of, I am sure DARPA & UNR have that on their Christmas list), so the only way to do it is via an omnidirectional pulse, which will fry anything not hardened.

                So, tricky to use, so not something most folks defend against (military does because nuclear weapons).Report

              • Michael Cain in reply to Morat20 says:

                The current big roadblocks I can see are AI limits (because, you know, jamming is a thing) and getting rid of the need for GPS.

                Back in the second half of the 1970s, my wife ate lunch regularly with people who wrote cruise missile software, and who would answer pretty much any question the pretty young woman asked. Even then, they did some amazing things with inertial navigation and terrain recognition radar.

                My favorite story she had from that time was one guy’s description of a recent test: “After launch, the missile flew down the valley properly, made its sharp left turn at the proper place, flew up and over the first mountain, flew up and over the second mountain, then flew smack into the side of the third mountain. Debugging this one’s going to be tough, given that all that’s left are parts scattered over several acres.”Report

            • North in reply to Morat20 says:

              Isn’t there a C&C problem for drones though? Couldn’t a couple jets being flown by a human kitted out with some kind of radio scrambler/jammer turn the swarm into scrap?Report

              • Troublesome Frog in reply to North says:

                It seems like that depends heavily on how simple the missions are. If they’re simple like “destroy this thing that was last seen at XYZ location and we don’t care if you come back or what other stuff you damage in vicinity,” that’s pretty tough to stop.Report

              • Morat20 in reply to Troublesome Frog says:

                Yep. Right now, the real limiting factor is how well they perform in autonomous mode. A problem currently being basically solved by a number of people trying to make self-driving cars.Report

              • Morat20 in reply to North says:

                Nope. You’d do it the same way you do missiles that target radar emissions.

                “Oops, I’m scrambled. Therefore, kamikaze the scrambling source”. After all, signal strength would grow if you’re heading the right direction, so it’s not hard to find.Report

              • greginak in reply to Morat20 says:

                It’s not like foes can’t counter that to some degree. Decoys are common. Heck a drone with a powerful emission source could make a potentially spiffy anti missile weapon.Report

              • Morat20 in reply to greginak says:

                Except, you know, drones are cheap. (Relatively).

                You jam a pilot the same way, only you’re limited in the number of pilots and very expensive aircraft they’re in.

                There’s always the evolving nature of the battlefield, but the thing about drones is — without all that massive amounts of equipment for keeping the squishy human alive, and the fact that performance is also limited to keeping the squishy human conscious, and of course being stuck reacting at slow squishy human speeds…

                You get better performance all across the board. For a lot less.

                So much of a fighter, or a bomber, or even a navy ship is based around the squishy human and their squishy human needs.

                What if you don’t need ejector seats, or a cockpit, or all those instrument displays, or reserve oxygen? What if you’re not limited to maneuvers that humans can stay conscious for, when your design doesn’t require “fitting a human and all his crap” into the machine?

                When you can replace the bulk of that with more dakka? Or make the vehicle smaller, more agile? Cheaper? Stuff in more fuel, more counter-measures, or just build a crap ton of smaller ones?Report

              • Oscar Gordon in reply to Morat20 says:

                Be on the lookout for the Navy to be deploying Lasers as point defenses very soon (they are under development right now).Report

              • Morat20 in reply to Oscar Gordon says:

                Last I checked, they had to keep the laser on target for a lengthy period of time (well, as these things go. A second or so) to do sufficient damage.

                Seems easily swarmed, or frustrated by things like rotation or reflective chaff.Report

              • Oscar Gordon in reply to Morat20 says:

                Depends on what you are trying to do. If you are trying to burn a missile out of the sky that is on terminal guidance, yeah, you’ll need a second or so of intense focused beam.

                But if you are just trying to permanently disable a drone, your attack can be very different.

                I mean, make no mistake, drones are a game changer, but we are a creative bunch of monkeys, we’ll figure out a counter.Report

              • Morat20 in reply to Oscar Gordon says:

                Well that’s warfare in a nutshell, isn’t it? Always a new offense, a new defense, a new tactic, a new counter-tactic…Report

              • Oscar Gordon in reply to Morat20 says:

                @morat20

                Uh, yup! Which is my point. Just because a new technology is on the scene doesn’t mean it’s the end of an era.

                When it comes to the Navy, what I strongly suspect is we are seeing the final days of the aircraft carrier as the core of the fleet. What we’ll see coming is SSNs*, CVNs & DDGNs along with littorals all sporting drone modules and each one maintaining a small wing of drones to do bad things to other people.

                *SSNs could surface, launch a bunch of drones from a topside bay, and submerge while leaving a small floating antenna array on the surface for drone C&C.Report

              • Morat20 in reply to Oscar Gordon says:

                Hmm. Combat’s gonna get weird. SSN’s fighting using swarms of autonomous drone torpedoes and anti-torpedo drones….

                I mean imagine a submarine going into combat starting with deploying a shoal of small, fast, one-shot little buggers around it whose sole job is to kamikaze those loud, incoming torpedoes. All you need is a shaped charge on the front end.

                You don’t even need that much speed on the things, because torpedoes will be heading right towards the sub (and thus the defensive shoal) so generating an intercept vector doesn’t have huge energy requirements….

                Of course after that, it deploys dozens of autonomous torpedoes that flood the area, trying to come in on novel vectors…Report

              • Oscar Gordon in reply to Morat20 says:

                See how much fun this is!Report

          • Maribou in reply to Oscar Gordon says:

            @oscar-gordon Ah, I see. I probably missed it because I, personally, had no idea what y’all were talking about :D. (Good to get the explanations… Now I’ve learned something.)Report

  4. Chip Daniels says:

    In time for Tech Tuesday:
    Elevators that go sideways.Report