0
$\begingroup$

When calculating large values with computers one is using Floating points, when calculating distances this can be a problem.

The floating point problem:

Eg. In early Kerbal Space Program the kerbins would hover above the ground 2-3 inches because of the floating point problem, this could be fixed either by calculating the distance from the nearest body or by upgrading to a 64 bit system.

Now in this world (made by a company, very happy for their trademark), the AI computers are really "zombified human brains" and it has access to large ships that can travel vast distances, by shifting in to another plane. Calculated and partially controlled by those brains or spirits, along with ancient holy computers.

Now, how big is the bit rate of the brain (after it has been clenched of personality) used purely for calculations and instructions, how "off" can one expect to be if by comparison kerbol's diameter gives a few inches inaccuracy, and those ships travels between suns.

$\endgroup$
  • $\begingroup$ the AI computers are really "zombified human brains" - Actually, it's only in MechJeb mod. And not zombified, but "mechanical copy of his brain". And, of course, not human but kerbal brain. $\endgroup$ – Mołot Sep 6 '16 at 11:04
  • $\begingroup$ KSP was to explain the Floating point problem, it has nothing to the story. The brains are machine spirits warhammer40k.wikia.com/wiki/Machine_Spirit $\endgroup$ – Magic-Mouse Sep 6 '16 at 11:07
  • $\begingroup$ I think you are asking what the performance of a human brain used as a general purpose computing unit would be? That stuff about 32bit or 64bit floating point numbers is really just a bug and not necessarily related to hardware (scientific applications for example routinely use arbitrarily large numbers, there are special software system for that). $\endgroup$ – Nobody Sep 6 '16 at 11:25
  • $\begingroup$ Or you are asking about accuracy of calculations? $\endgroup$ – Nobody Sep 6 '16 at 11:29
  • $\begingroup$ @Nobody the floating point bug is because rounding, so if i know what the aproximate bit value of a brain is i can reverse calculate the rounding error and get the inaccuracy. $\endgroup$ – Magic-Mouse Sep 6 '16 at 11:43
3
$\begingroup$

I think the error of your space travel will be much more limited by the physical instrument used for measuring, rather than a (hypothetical) bit brain.

  1. When calculating large values, one can use machine floating-point (with differences in precision between 32 bit and 64 bit). This is absolutely not the only way to compute. One could use arbitrary precision floating point (why not compute on 1024 bits?), which does not require your CPU to have that same precision.
  2. While a 64 bit floating-point basic operation gives approximately 10e-10 error, most physical instruments used to measure do not achieve such precision.
  3. Most crucial floating-point errors are due to not taking into account that there are floating-point rounding errors. When this is taken into account, algorithms, mathematical formulas and models can be used to measure that error, and determine the maximum error that could occur from floating-point computation.

Hence, the real problem would occur from the error on measurement. If you measure the angle between two stars with a 10e-5 precision, that would leave millions of kilometers of error in the distance between the two stars, if they are far enough. (I did not do the actual computation, which is however easy to do)

Once you know your physical measure error and physical formula used, then you adjust your floating-point precision used to be negligible comparing to the error measure.

Of course, to safely travel in space, you would need a physical modeling that take into account those measurement errors. I'm sure this is also done in real world space travel, and even in plane flight.

$\endgroup$
  • $\begingroup$ Also, I think that neurologists mainly think that there is quantum physics involved in the brain computing/consciousness mechanism. (source needed) It would probably indicate that the brain computes more than a computer (not necessarily faster, but a different and larger set of problems), since it would not be equivalent to a Turing machine. Which means that the "brain equals computer" metaphor is invalid, and brains have no bits. $\endgroup$ – DainDwarf Sep 6 '16 at 11:56
  • $\begingroup$ Larger set of problems? I've never looked into it too much, but I got the impression that it's mostly a very different set, with some "basic" (from a classical computer scientists point of view) functionality lacking from quantum computers. $\endgroup$ – Nobody Sep 6 '16 at 12:06
  • 1
    $\begingroup$ Sorry neurologists don't think quantum mechanics is involved in the brain's mechanisms. So no source is required. A few outlier non-neurologists may think so, real neurologists think they're wrong (to put it politely). Definitely the brain computer analogy is bunkum. $\endgroup$ – a4android Sep 6 '16 at 12:07
  • 1
    $\begingroup$ Also there is the fact that you don't have to use 'floating-point' numbers (depending in your programming language [what language do you program zombie-brains in anyways?]), there are other data types that will go to as high of a precision as you need. $\endgroup$ – Marky Sep 6 '16 at 15:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.