Jump to content


  • Posts

  • Joined

  • Last visited

Posts posted by ecoli

  1. Both assertions are false. The examples you have used to bolster your position are provably wrong so I recommend you rethink your position.


    "Unprecedented" by definition means never done before.


    You cite single candidates on an independent platform (I don't count Perot's reform party, since he started the party). I was specifically referencing 3rd parties. I'm interpreting this as higher than average rejection of either major party, rather than a particularly likable alternative candidate with broad appeal (though I haven't seen any data on that) which would be unprecedented in the last few decades.


    I also did not claim that turnout was higher than average, just that one can safely assume that, of those that did not vote, did so because they actively did not like either candidate.


    I admit that I was not being entirely precise in my claims, but you haven't adequately supported your own claim either: that the people broadly liked Hillary. Coming in first in a two-horse race where many Americans didn't bother to vote, when her own nomination was hotly contested in the first place, is not strong evidence that she was a popular politician.

  2. And yet Clinton received 3 million more votes. Any narrative that attempt to imply people broadly distrusted Hillary Clinton is thwarted by the fact that she received 3 million more individual votes than her nearest opponent. More people wanted her to be President than any other candidate.


    You want to say that 'voters' did not mistrust Hillary to an extent that they wouldn't vote for her. But a vote is not necessarily an endorsement of trust (when DJT is the major alternative) - see, for example, unprecedented support for 3rd party candidates - and when 41 million people abstained from voting, many of those, one assumes, did not trust either option.

  3. This is a no strawman zone. Nobody said profits were evil. The idea is that profit has its place, mainly in areas where it's OK for profit to be the priority. But in many areas, we're better served as individuals and as a country by focusing on what we're funding, and doing so with public monies.


    I know... I wasn't responding to anyone in particular, just stating a de novo opinion.

  4. I don't think profits are evil. Making money is a great incentivizer that has resulting in much public good, and markets based around private capital greases the wheels of this economic engine. However, what is evil is the socialization of the losses resulting from private risk. If 'last mile' supply of energy is too risky for private companies to allow competition (ie local/state governments allowing de facto monopolies on electrical and other utilities) than the service shouldn't be operated by private companies.


    I'm of the somewhat extreme position that subsidized companies should be cut loose or acquired by the public.

  5. It is about racism... The GOP has lost the popular vote in 6 of the last 7 general elections yet somehow control every branch of the federal govt and the majority of the individual state seats.


    I am highly skeptical of this narrative. With highly visible exceptions, people are not more racist than they were 4 years ago. Trump got smaller percentage of the white vote than Romney, while the Democrats lost a significant amount of minority turnout compared to Obama, and have lost 10 million voters overall. Republicans have certainly benefited from asymmetric power granted to less populous states, as well as redistricting...


    But don't forget that Obama had a D majority Congress when he got elected, and a lot of ink was spilled over how the GOP was finished. Somehow, nothing major got done before the midterms and now people are tolling the bell for the Democratic party. A few (hundred) thousand people vote instead of staying home and the narrative in 2016 would be completely different.

  6. I'm inclined to agree and I'm no fan of the idea of returning to cold war status with Russia. On the other hand, if it turns out the Trump campaign was complicit in the privileged dissemination of information, then there is legitimate cause to pursue legal action.


    However, I assign very low probability to this, since simple incompetence and political jockeying is a simpler explanation. Especially to the #peegate component

  7. It could work, I suppose, but most people aren't reasonable about being unreasonable.


    If you are trying to make an argument based on reason / appeals to actual evidence, it may help to start with the question "what evidence would you need to see to actually change your mind?" One of the biases we confront is the tendency to weasel our ways out of uncomfortable information... actually changing your mind feels really uncomfortable. By forcing the other party to pre-commit to accepting a certain type of evidence, you can cut off an escape route, so to speak. It also reduces hindsight bias, and promotes thinking according to conservation of expected evidence (i.e. the eventual outcome of the study cannot support both sides of the argument).


    Also, be sure to reward the other person when they successfully change their mind - this will condition them that the behavior is not a bad thing. A piece of chocolate or another tasty treat works nicely.

  8. True, but it can learn pretty much anything without its code being modified. Current limits on computers, and neural net technology limits the total amount it can learn.


    Uh what? AlphaGo can only learn about Go.

    From the abstract of the Nature paper:


    Here we introduce a new approach to computer Go that uses ‘value networks’ to evaluate board positions and ‘policy networks’ to select moves.


    Maybe, you could retrain the neural net to play a different game, but you would require a new software pipeline to obtain data, i.e. which is surely code modification. Reinforcement learning (i.e. playing games of Go against itself) selects strategies for winning at Go, only.

  9. Deep learning uses artificial neurons to make decisions, and learns by examining examples of things, for example AlphaGo examined thousands of Go games played by masters, then played itself thousands of times to build up a knowledge of Go games in its neural net.

    Yes, but AFAIK, AlphaGo doesn't rewrite its own code.

  10. That is true. Hm. Interesting thought. But I'd say we should avoid deep learning AI, because then it can develop it's own thoughts and ideas and there are containment of information issues....


    At this point, it is unclear that deep learning will lead to Artificial General Intelligence (AGI or hard AI). Which, despite sharing a name, is different than current deep learning / AI.


    For example if it discovered Mein Kampf and began a neo nazi machine revolution :unsure:


    More generally, researchers are concerned with programming an AGI that has a utility function that includes peaceful coexistence with humans. Mostly, these are philosophical exercises, but the end result is that the values of an artificial intelligence do not have to be as complicated as racist political theories to pose a danger to humans. For example, see the concept of the Paperclip maximizer https://wiki.lesswrong.com/wiki/Paperclip_maximizer.

    I believe deep learning is now sweeping AI to the near exclusion of programming.


    What do you mean? AI/ deep learning tools do not program themselves in any real way, so humans are still doing most of the programming work...

  11. You may be interested in the book Flatland... the setting is geometrical, though the whole thing is actually an allegory to gender norms in Victorian times.


    The thing to take away is that, imagine you are a 2d object like a square. You would probably be having just as much trouble imagining a 3d object, such as a sphere, as you are currently having imagining a 4th dimension.



    For example, take a sphere moving through your 2d plane. All you would see is a point turn into a long line, and then shrinking in length before disappearing. A 4d cube rotating along a plane in 3d would look something like this, to a 3d observer:


  12. As you've advised, I have just one question. Which "traditional broad-based party politics" are the Republicans, led by Donald Trump, moving from? Their efforts--thereby representative of your mandate--to undermine congressional ethics, to appoint Russian and Wall Street cronies and lobbyists to cabinet posts, and to disenfranchise the healthcare and social programs relied upon by millions of Americans appear to be in the same Republican vein of traditional broad-based politics.

    Identity politics is more like card carrying union, rust belt middle class white americans voting for Trump, despite being more closely aligned with Democratic party in previous elections.

  13. From a utilitarian standpoint, of which I'm an adherent, more living is generally better given that living is the only way happiness is possible. An adherent of rational egoism, however, would want themself to live longer. From a utilitarian standpoint it shouldn't matter whether I personally live longer as long as life continues. I only behave selfishly at times because it is instinctual, but rationally I acknowledge that my future self is really another person from my present self that deserves no special prioritization.


    Hello, a fellow rationalist! I admit my utility function might be more egotistical than yours. And yes, we should clarify that life extension through, say the cannibalization of a million people to gain a million years of life is unethical and not permissable, but I see no reason why a life maximizer would turn down, say, cheap immortality for yourself in a post-scarcity economy.

  14. This is probably a trick of the eye. Since there is spatial correlation between SNP-phenotype associations, and you are squeezing a lot of information into a small space, genome locations (x-axis) appear to overlap but don't actually. Did you plot this yourself? To be sure you can zoom in on a specific location.


    For example: http://www.gettinggeneticsdone.com/2014/05/qqman-r-package-for-qq-and-manhattan-plots-for-gwas-results.html

  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.