• 50 Posts
  • 1.47K Comments
Joined 11 months ago
cake
Cake day: August 15th, 2023

help-circle

  • Regardless of the side, if the convicts have proper training, supplies and aren’t used as pure cannon fodder I don’t have any issues with it personally.

    Honestly, I am surprised that Ukraine wasn’t doing it sooner. Unfortunately, the front has been somewhat slow for the last year or so (with an exception or two) and that means both sides are dug in fairly deep.

    If Ukraine decides to push back harder and start retaking large portions of land, their losses are going to skyrocket. Regardless is you agree or not, using convicts to push the lines makes sense and doesn’t risk better trained troops.

    That is just how wars of this type work and the offensive side will need to absorb more damage.

    Personally, if I had the option of fighting or rotting in prison, I would fight and would probably volunteer for some of the hardest work. The thing is, it would need to be my choice and I hope that any convicts, Russian or Ukrainian, get a choice in that.









  • you’ll be doing the QA and will in fact be working on the parts / product to get it to where you need it to be.

    Absolutely. Unless a person wants to spend thousands of dollars on push button solutions that cover every imaginable use case, customization is the way to go.

    For solid machines, the customer should already have an idea about what parts need to be modified. If a machine was advertised to mill a widget at +/-20% tolerance, cool. If you want to spend $500 more on a custom pully to get withing 5%, awesome. Precision is expensive and customization is niche.

    For cheap machines, everything is generally ravaged by bean counters at every level of design and manufacturing. As long as people understand this and can make repairs, that is sometimes OK.

    While I feel OPs pain of finding a 2¢ part that was 0.3mm off center, I can only just shrug it off. A pseudo-premium 5¢ part or building a jig for a worker to test each gear would have been quite expensive and it would probably tack on $2-$5 to the end product price. ($2-$5 actually matters on sites like Amazon or Temu and could potentially cost thousands in lost sales due to product placement.)




  • That is a photoshopped picture, btw.

    I can’t imagine that keyholing like that would be even be slightly tolerated and it’s not too difficult to get stable flight unless there are serious issues with the firearms.

    If a bullet is tumbling out of the barrels, those barrels are seriously fucked. They are shot out, are using bullets that are too small or there are serious defects with the bullets.

    A bad twist rate or extremely poor load generally won’t cause tumbling until much later. If the bullets start at slightly at supersonic and slow to subsonic quickly, that might do it.

    I would like to see those videos of the keyholing you described, which is why my brain seems to have vomited all over this comment. (This has got me super curious, is all.)




  • That sounds about right. QA is expensive and time consuming, so it’s left up to the customer. This applies to every single part in the supply chain.

    If you want a set of mitsumi linear rails for real precision applications, it’s going to cost just as much or more than that printer.

    This is not a “buyer beware” rant, but the buyer should know they aren’t paying for consistency or precision. I am basically saying that for these printers to work reliably and with proper precision, you need to tear them down yourself and inspect each bolt.

    I buy cheap Chinese stuff all the time, but my process is to tear the product down and find where costs were cut and look for any serious dangers.

    Svol is well known enough that you should be able to get replacement bits for free. Or not. It’s a crap shoot, TBH.



  • I have been to some interesting places in my own brain with the help of psychedelics. While I have experienced different levels of self-awareness, the possibility of that “filling up my brain” is likely not possible.

    There is a part of the brain that psychedelics specifically affect that functions as kind of a traffic regulator. It typically only directs signals from one part of the brain to another part. Psychedelics open these pathways up and allow for information to flow to in all kinds of directions. (Synesthesia, sensory confusion, is an example: Feeling colors and seeing sounds.)

    What the experience does for me, is that I seem to gain more awareness into how my brain works. It’s like I can stand back and watch how my brain processes things. My subconscious is pulled into full view and I can almost tinker with it, in a way. Some theories suggest that the colors and geometric patterns people tend to see is our actual brain operations being leaked into the visual cortex and/or the fractal patterns are the result of actual “data loops” caused by psychedelics allowing information to pass around freely. (Not my theories, btw.)

    Now, you may read this and think: That dude is just tripping! (And you would be very much correct.) The thing is, every experience I have and anything that I feel is already in my brain. The data is already there, but how that data is processed is vastly different. Even if I am perceiving parts of my brain that I couldn’t before, it’s still just the same neurons that were always there.

    So, what I am basically saying is, is that my self-awareness temporarily becomes self-aware. It’s a shitty description, but it’s the closest I could get to matching the situation you were asking about.

    I still develop memories, almost as normal. Some memories stick and some fade. All that really happened is that a few neuron weights got shaken up and it all becomes a few new pathways with a similar number of neurons as before. (Neurogenesis and dendritic growth as a result of psychedelics is a different subject and I wouldn’t think it would be part of the recursion-type situation you are asking about.)

    Memories become integrated with existing ones, basically. While vastly different in many ways, our current AI tech has a set number of “neurons” it can work with in any given model. You can train those same bunches of neurons with millions of different patterns, but the actual size of the neural network doesn’t change, no matter how much data you cram into it. With slight changes to when neurons fire, you are using specific pathways of neurons to save data, not necessarily using the neurons themselves.