He sounded like he was deliberately avoiding a name, but I guess implied meanings are different over the internet, and I have no idea if I'm missing some background.
So... the elephant in the room? The nomination phase is almost over, and we have no candidates. Shouldn't we be concerned? What happens if nominations close and no one decides to run for moderation? Can we, the rest of users, nominate someone else and hope they'll agree?
Being a mod would actually take away some of the freedoms I enjoy now.
For example I can vote to close fairly freely knowing that four other people have to agree with me for anything to happen. As a mod I would be very hesitant to VTC except in the most egregious cases.
I mean you're already most of the way there. All the privileges unlocked, room owner powers, well known in the community, etc. Honestly the only difference is you'd be an official mod :P
Quick Update
We welcome the reception this post has gained, and appreciate all of you who are considering to run. We will therefore hold an election. As soon as new details are available, they will be featured in a separate Meta post.
Again, thank you all for your participation.
This communit...
@JohnRennie I would be terrified of automatic mod powers. Is it you who I'm thinking of who is usually really reluctant about closing with your automatic close badge?
@JMac the dupehammer, yes. I will often vote to close as Off topic/Other and type a Possible duplicate of message to avoid immediately hammering the question.
I mean that said you can probably recognize when a question needs closing as well as any other mod can, given you've been here for longer than most people on the site
@SirCumference The thing is that I believe strongly in the site as a resource for budding young physicists. My going in point is that you should be able to use info from any answer of mine in an exam and get full marks.
... I wonder if it would be a terrible thing for a mod to use an alternate account for regular flagging. That seems like it would be a really legitimate use of secondary accounts
@JMac Sure. And on the other hand, moderators must be willing to use the tools they're given, and they will get attacked for it. (aka "You aren't a mod until a user you just took action against has cast profane aspersions on your entire ancestry in response")
If someone thinks defaulting could be disastrous, let me remind you that users that have been suspended during the last year cannot nominate (unless there are extenuating circumstances discussed in advance with the SE team).
At any rate a 1 option moderator isn't really the best idea imo. It literally hands power to someone without anyone agreeing on how they'd affect the community
@ACuriousMind Okay, so the panic button nomination strategy would have been the actual solution there I guess. I thought that worst case scenario, I'm sure a few people would have popped up had that situation happened... but still seems strange that nothing could be done
I described SE elections in the last election already as an elaborate game of chicken: Given the choice between 1. another highly qualified user becomes moderator and 2. oneself becomes moderator, almost everyone prefers 1. So most qualified users won't nominate until they are convinced that no one they consider more qualified than themselves will nominate
@EmilioPisanty The last election had a lot of nominees (and non-nominees) who had discussed their strategy for the chicken game already in advance in this chat ;)
@JMac Not explicitly, but for instance I partly nominated so early because I already knew that many people I'd really liked to have seen running wouldn't run at all. There were also several, ahem, low-quality nominations (that were later removed) that I think sort-of forced early nominations from other users
@SakshamRastogi Probably because of the feedback to their questionnaire answers, but that comment is all of the public statements we've got on that matter.
Re this question: am I correct in thinking that only 2D (or 1D) matrices multiply is the way we normally multiply matrices. For higher dimensions we can take a tensor product but that's a rather different thing.
Yeah, for tensors of rank higher than 2 there is more than one possible way to "multiply" them - it's called contraction over different tuples of indices, then.
I guess that's the problem. The work matrix strictly means a two index object. Anything with a larger number of indices isn't a matrix so asking what matrix multiplication means in that context is meaningless.
@enumaris Let's try: The H-C correspondence is the idea that asking whether a formal type is inhabited (i.e. there exists at least one object of the type) is a problem of exactly the same kind as asking whether a logical proposition is true.
They need to have "algebraic types", which at the very least means they need the notion of a product type (a tuple) and a sum type (an either-or type, like a union in C, or an enum in Rust)
The product type and its correspondence is probably the easiest to understand because the tuple is a pretty common structure.
Examples are really the key to demystifying the correspondence. Let's take the logical statement "A and B". Then it is true if and only if both A and B are true.
The corresponding statement about types is that for two types A' and B', the product type that describes tuples of one A' and one B' has at least one inhabitant if and only if there is at least one object of type A' and one object of type B', because otherwise you can't construct a tuple
@enumaris It's a pretty handy thing. Given two types S and T, it's a type that says "I'm either an S or a T"
To discuss the full correspondence, in particular what it does for "for all" and "there exists" quantfiers, we'd need to discuss some stranger types which you usually see as "generic types" in programming languages, if they exist (they occur e.g. in C++ templates, parametrized Rust types). I don't think Python has them either, because Python hides its type system from the user and I don't think it has use for generics in this sense, but I'm sure @PM2Ring will set me straight if I'm wrong
(To be clear: Very formally neither C++ nor Rust actually has these types, since they are made "ungeneric" (monomorphized) at compile time. but I don't know any other non-functional languages that have these types, and functional languages like Haskell are somewhat dissatisfying examples because they are explicitly built to have a type system for which the H-C correspondence works perfectly.
That is, these languages are in a way better examples, but they don't serve someone very well who doesn't yet understand the more abstract notions of types
Python isn't helpful for modeling this kind of stuff. In most languages, a variable is essentially a box for holding a value of a specific type. So you cannot assign a value to a variable if the value doesn't have the type that the box was defined with. Although most such languages will perform safe conversions automatically, eg if you assign a 8 bit integer value to a 32 bit integer box. In contrast, in Python, a variable is just a nametag, with no intrinsic type. A Python object may have 1 or more nametags, or None, and its type is an intrinsic property of the object itself, not its namet…
@enumaris Sorry, I must retract my statement just there. The generic types I was thinking of are not enough for true dependent types, and Haskell doesn't have them either.
@enumaris Haskell will blow your mind. :) But for practicality, I advise you to learn C solidly first. I tried learning Haskell a few years ago, and although it was interesting, I found some of its concepts hard to grok. But I might give it another go...
@enumaris The types that correspond to the logical quantifiers are so-called dependent types. I briefly fooled myself into thinking that generic types would be a close enough analogy, but they're not.
I think it's helpful at a high level. Can't say I have a great understanding right now, but that's probably just cus I don't have the background necessary to really understand it well
I can relate to that. OTOH, I learned quite a few languages for a few decades before I learned Python. It's good to learn a few languages of different language families, just to get exposure to the various different ways that those families do stuff. Even if you end up doing most of your coding in Python, the awareness of those other languages will enhance the way you approach a coding problem.
Coding in C gives you more direct control over the machine, and the feeling of power from being so close to the metal is kinda nice. :) And the fact that code written in C runs 100 times faster than pure Python's rather nice too. Of course, Python often leverages the power of C, eg a lot of list operations can do most of their work at C speed. And for serious number crunching we can use Numpy, which does most of its stuff at C speed.
@SirCumference I was convinced into running. Kind of glad I didn't win. And while I'm usually always around these days, I'm not attentive enough to make a good mod
> 'a, An iodine-124 atom can decay with a half-life of 4.2 days to an atom of tellurium-124, through a process called electron capture. The nucleus of the iodine-124 atom captures an electron from the electron shells that surround it. A proton (circled) in the nucleus is converted into a neutron, and a neutrino is emitted. b, A xenon-124 atom cannot decay by electron capture, because of the law of energy conservation.
> However, it can decay with an extremely long half-life to a tellurium-124 atom, through a process known as two-neutrino double electron capture. The xenon-124 nucleus captures two electrons from the surrounding electron shells, which results in the conversion of two protons (circled) into neutrons, and the emission of two neutrinos. The XENON Collaboration2 has measured the half-life of this process to be 1.8 × 1022 years — about one trillion times the age of the Universe.'
The opening paragraph is incredible (bold font is mine)
> 'For half a century, our view of the world has been based on the standard model of particle physics. However, this view has been challenged by theories1 that can overcome some of the limitations of the standard model. These theories allow neutrinos to be Majorana particles (that is, they are indistinguishable from their own antiparticles) and predict the existence of weakly interacting massive particles (WIMPs) as the constituents of invisible ‘dark matter’ in the Universe.
> Majorana neutrinos mediate a type of nuclear decay called neutrinoless double-β decay, an example of which is neutrinoless double electron capture. A crucial step towards observing this decay is to detect its standard-model equivalent: two-neutrino double electron capture. In a paper in Nature, the XENON Collaboration2 reports the first direct observation of this process in xenon-124 nuclei, using a detector that was built to detect WIMPs.'
Is it possible for consciousness to be just a genetic trait? Also why aren't dolphins smarter? Does the brain size matter for intelligence and consciousness...? Is it possible to raise a dolphin like a child
@NovaliumCompany I certainly don't have a full overview of what has been done in ML, but I have difficulty seeing what the problem here is at which machine learning is supposed to be better than traditional algorithms.
Oh-oh. "A PMH is the simplest confirmation of -- what we've mistaken to be a separate force known as -- gravity generated by the magnetic lock of a hollow planet." From physics.stackexchange.com/a/476189/123208 The original version of that answer was relatively innocuous, but the sneaky SOB edited it 2 hours afterwards.
All of these mathematics, physics and so on are just looking through the key hole and interpreting what we see, we call it through and now when we have the chance to open the door, people are scared and want to run away.
AI is coming. We either prepare, or go extinct. Both ways, it's still technically evolution.
Morality is relative, so that's not a valid argument.
It may be a scary future, I agree, but my curiosity is stronger than my fear. And if I've learned something about fear, it's that some of the time if protects me from real dangers, but most of the time it just blocks me from seeing and doing more.
@NovaliumCompany Maybe it is, but nobody really knows how to create an AI. But some interesting stuff has been learned, and some useful stuff has been made in the process of working on AI. Eg, natural language processing has made a lot of progress. But we still don't have perfect machine translation because a good translator needs to understand what they're translating, unless the material is fairly simple.
But I'm concerned that someone might develop powerful AI before we've discovered how to make friendly AI. I'm no disciple of Eliezer Yudkowsky, but I agree with him that it would be a calamity to develop a true AI that wasn't friendly.
@NovaliumCompany there's actually a whole branch of AI research that's based on trying to imitate the cognition of the human brain. It generally stands in contrast to the standard data-driven ML methods...but bringing the two together is an active area of research.
welp, we are now officially a generic "AI" innovation center. No more healthcare domain...whomp whomp...
my research plan was written with the healthcare application in mind tho...so now I'm not sure what to research lol
and this is also why I'm not a big fan of documentation....produced 17 pages of documentation...research plans on how to go forward building the application...and then the project is canceled and now those research plans are worth dick
@enumaris (sorry to hear of the disruption, but...) the plot thickens! maybe mentioned this before, curiosity is a hot/ rising topic in ML/ AI research... can help you out with lots of juicy leads if you go in that direction :)
So... the elephant in the room? The nomination phase is almost over, and we have no candidates. Shouldn't we be concerned? What happens if nominations close and no one decides to run for moderation? Can we, the rest of users, nominate someone else and hope they'll agree?
Quick Update
We welcome the reception this post has gained, and appreciate all of you who are considering to run. We will therefore hold an election. As soon as new details are available, they will be featured in a separate Meta post.
Again, thank you all for your participation.
This communit...
