31 Matching Annotations
  1. Feb 2024
  2. Sep 2019
    1. Merrill’s utility based substudy is suspicious be-cause it was unable to detect the fact that, e.g. 2-candidate majority vote is non-optimal from a util-ity standpoint, i.e. has nonzero Bayesian regret.(All his data for 2-candidate elections had “100.0%social utility efficiency,” in his terminology.) Thatsuggests that Merrill’s computer program had bugs.

      This isn't due to any bug. Merrill's study uses normalized utilities, so in the 2-candidate case, the majority vote always goes to the utilitarian winner.

  3. Aug 2019
    1. the standard AES17 dynamic range measurement

      AES17 doesn't define "dynamic range". It defines "Signal-to-noise ratio or noise in the presence of signal" which is what includes the test tone:

      The test signal for the measurement shall be a 997-Hz sine wave producing – 60 dB FS at the output of the EUT.

  4. May 2019
  5. Feb 2019
    1. Any election system that favors extremists would be considered unreasonable; the same rationale must be applied to moderates.

      Utter nonsense. To paraphrase:

      Any election system that favors unrepresentative candidates [like IRV] would be considered unreasonable; the same rationale must be applied to one that favors representative candidates.

      Uh, no. That doesn't follow.

      FairVote starts from the conclusion that IRV is the best voting method, and then works backwards to try to justify it, in this case arguing that a flaw of IRV is actually a feature, by making a false equivalent between a voting system that favors unrepresentative candidates and one that favors representative candidates.

      The whole point of an election is to find the most-representative candidate.

    2. Agreeing that the Condorcet criterion is desirable is equivalent to saying that moderate candidates should always win.

      Yes, candidates who are moderate relative to the voters should always win.

      The goal of an election is to find the candidate who best represents the electorate. If the electorate is left-wing on average, the winner should be too. If the electorate is "strong on both personal freedoms and economic freedoms", then the winner should be too.

      Anything else is undemocratic.

  6. Nov 2018
    1. Under range/score, the best strategy to promote the election of a preferred candidate is always to give that candidate the maximum score and then give every other competitor the minimum score.

      Yeah, this is false.

      If you have perfect knowledge of how everyone else is voting (and you usually don't), then the best strategy is to give a maximum score to the frontrunner that you prefer, and also to everyone you like more than them, and to likewise give a minimum score to the other frontrunner, and to everyone you dislike more.

      This is not bullet voting; it's equivalent to Approval Voting, and leads to more moderate winners who are good representatives of the electorate.

      Real-world Score elections don't show this behavior, anyway, because polls are imprecise and the consequences of voting honestly under Score aren't as dire as they are under FPTP or IRV.

    1. On the theoretical front, approval and score voting fail a critical test that voting theorists call the majority criterion.

      "Failing the majority criterion" is actually a good thing, because these systems find the candidate who is most acceptable to the entire population, rather than the candidate who is most acceptable to half of the population.

      Majoritarianism that ignores the desires of half of the population leads to adversarial politics, polarization, and eventually even to violence and civil war.

  7. Oct 2018
    1. If the SM58 noise floor is calculated at room temperature, the voltage output is 0.00000032 volts.

      This is equal to -130 dBV. The SM58 Vocal Microphone Specification Sheet says:

      Sensitivity (at 1,000 Hz Open Circuit Voltage) –54.5 dBV/Pa (1.85 mV)

      At this sensitivity, the self-noise would be 94 - (-54.5 - -130) = 19 dB SPL, which is pretty typical and certainly not "lower than can be typically measured".

  8. May 2018
    1. plus the impedance of the path from the inverting input to ground i.e. R1 in parallel with R2.

      This is incorrect. If R1 is infinite and R2 is 0, the parallel impedance is 0 ohms, but the input impedance is much higher than the input impedance of the op-amp itself, due to feedback making the inputs very similar in voltage.

      The input impedance is actually

      $$(1 + A_0 B)\cdot Z_\mathrm{ino}$$

      where

      • \(A_0\) is the gain without feedback (the open loop gain)
      • \(B\) is the fraction of the output which feeds back as a negative voltage at the input
      • \(Z_\mathrm{ino}\) is input impedance without negative feedback

      For the above buffer example, it would be close to \(A_0 Z_\mathrm{ino}\)

      See https://electronics.stackexchange.com/q/177007/142

      Simpson - Introductory electronics for scientists and engineers section 7.2 Negative Voltage Feedback explains this clearly

  9. Apr 2018
    1. 10–23)

      Equation 10-23 is incorrect:

      • The noise of the resistors does not sum, since the resistors are in parallel and load each other. It would be equal to a single 50 kΩ resistor producing 4.0 μV.
      • The 113.1 value is output noise. The resistor noise needs to be multiplied by the gain to get the output noise, too.
  10. Mar 2018
  11. www.analog.com www.analog.com
    1. The poles of the Bessel filter can be determined by locating all of the poles on a circle and separating their imaginary parts byn2where nis the number of poles.

      This is incorrect:

      • The poles are not located on a circle
      • The imaginary parts of the poles are not equally spaced

      To generate the poles of a Bessel filter you need to use root-finding methods on the reverse Bessel polynomials. There's no other shortcut that I'm aware of.