Hacker News Reader

Comment by j16sdiz | Hacker News Reader

This sentence captured my mind:

> Scientists still debate whether viruses are alive—they can’t replicate independent of a host cell’s molecules—but there are even simpler “creatures.”

We have varies "definition" and "categorization" saying virus is not "alive".. you can see many biologist saying this definition does not capture the essence and need changes... There are published paper for that every few other years.

Look like this is what the "AI is sentient" thing going... We will get a textbook definition for sentient AI, but nothing could end that debate.

View on HN

My background is philosophy of language. The vagaries of definitions are exactly how many get communication backwards. It’s jut important to remember that language is effectively just pointing at things, and that the words we use are effectively arbitrary.

The distinctions in taxonomy should be reflected by the language, and we should very much not try to conform our taxonomy to our existing terms… beyond, say, shorthand.

This is clear to many of us with no formal training in language.

I completely agree, and wouldn’t suggest otherwise. I’m mainly referring to the concept of platonic ideals, which seem to be at least somewhat instinctual.

Yes, that was my original understanding of your meaning. There are other experiences and traditions as well that illuminate this way of thinking.

  we should very much not try to conform our taxonomy to our existing terms

I see this stunting imagination everywhere. Economics, science, nature, etc. Humans generally feel they've "captured" knowledge with a brief overview and disdain any messiness that might allow that knowledge "to escape".

Some humans are curious to be exploratory of knowledge that "escapes" these classifications, as you say :) this is, in some definition, the criteria of an artist of sorts. Maybe artistic minds are just scientists of another cloth, investigating the features of linguistic creatures that escape the boundaries that other (more "particular") types of minds create and impose

This is maybe the most beautiful description linking the arts and sciences that I've ever read. A good scientist is one who extends the envelope of our external knowledge; a good artist is one who pushes the boundaries of our inner qualia and interpretations

>It’s just important to remember that language is effectively just pointing at things, and that the words we use are effectively arbitrary.

I think the success of transformer-derived LLMs suggests that Wittgenstein was more correct about contextual “family resemblances” than the “pointing at things” school of language philosophy.

I'm a bit confused by your statement.

I'm a bit ignorant of the actual math/logic for modern LLMs (I took Andrew Ng's course on machine learning, but that just gets you to basic neural network architecture).

I'm also confused as to why you would put Wittgenstein in opposition to the concept of language as "pointing at things," as I would put him very much in that school. I'm a bit rusty, but I've been through the Philosophical Investigations a couple times, and I'm almost certain he's opposed to platonic forms.

He thinks some words rely on ostensive definitions, colour for example. Otherwise, though, his view of language can roughly be summarized as "meaning is use". You're correct to read him as opposed to platonic forms and ideas; however, he argues that language and meaning has its roots in behaviour, or as he puts it: "a form of life". So for him meaning is relative but not arbitrary. It's a crucial distinction.

I only wish I could have heard him lecture. His two books, while interesting, have always seemed, well, a bit odd in their structure (less so with PI). Beyond that, that people use some random notebooks like his published texts also seems odd.

People like definitions that have clearly defined boundaries but biology is not like that.

There can be a messy continuum between is and is not. Things can match a definition strongly by some measures but miss some that are critical to how other things meet the definition. At best, a definition is a constellation of semi-related characteristic where some organism meet some subset and other meet another subset, but there isn’t a 100% match and may not even be an overlap.

> There can be a messy continuum between is and is not.

Exactly--concepts like "life" were invented long before we had an inkling of what viruses were...is it surprising that it's not a perfect abstraction for what we find in reality?

I think people have no qualms about fuzzy boundaries or continuities. The problem arises when those get subsequently mapped onto discrete sets with wildly different consequences.

My work in computational ontology led me to believe people are fast to come to a conceptual joint agreement in a given model when the outcome is not consequential, but find it nearly impossible to settle when there are strong implications attached to a labeling.

Language only works between people that ultimatly want to agree.

If labeling has fuzzy boundaries, perhaps it shouldn't have strong implications. If it does, the boundary cases behave chaotically, which is usually not acceptable behaviour. I mean, if the labels have no consequences, what is the point of labeling in the first place?

Organizing everything in the world into neat boxes feels so early 20th century to me. Back when we still thought that human reasoning can be applied to achieve a complete understanding of the world and everything in it. Putting things into boxes is a coping mechanism for the feeling of chaos; people do it because it makes them feel good, not because it actually reflects how things are.

>If labeling has fuzzy boundaries, perhaps it shouldn't have strong implications.

Well, it's tricky because labelling is so useful (oftentimes crucial for communication purposes), but we would be hamstringing ourselves if we either a. held off on labeling things with fuzzy boundaries altogether or b. refused to entertain possibilities with strong implications because a labelled phenomenon has fuzzy boundaries.

Best to remember that labels are useful but with very important limitations. Ultimately, when we get out to the edge of our understanding the boundaries will always be fuzzy, at least until a breakthrough (and sometimes long after). Whenever you are at a boundary, continually remind yourself that the label is less important than communication and exploration of what actually appears to be going on.

Are viruses alive? This is a very interesting question to try and answer from different angles, but arriving on a definitive "yes or no" here is not at all the most important thing about viruses to be worrying about.

We label because we need to compress reality before we can reason and make sense. Is this a limitation of human intelligence? Sure, but I would hypothesise it is even more fundamental. Irreducible complexity can only be observed and reacted to, it takes away the proactive. So we make do with reduced models that are definetly wrong, but sometimes usefull.

In a way, a virus is not too far off from a complex prion. Simple single cell organisms aren't too far off from complex viruses.

There's a line of thought that metabolism is the defining feature of biological life, and viruses do not have any metabolic activity of their own. This is essentially the only real distinguishing characteristic that holds true between viruses and organisms.

Once a virus has a host cell, however, it uses metabolic events to do its thing. That use of those events to drive its own reproduction is very life-like, and is very unlike a prion. In terms of behavior, it really is more like life than not. It uses metabolism, even if it doesn't innately perform glycolysis like everything else.

The "are viruses alive" is really more a medical philosophy question than a practical one.

And then on the other side of prions, you get things like asking whether self-catalyzing crystal formations are alive.

In what way is a virus remotely like a prion? Both the structure and function of the parts are completely different.

Just because a virus interacts with a metabolism doesn't mean it's closer to having one of its own. That's like saying humans use gasoline combustion for our biology because we drive cars instead of walking sometimes.

You're not wrong that whether viruses are alive is not a very practical question, but none of the rest of your post meaningfully addresses it either way.

A viral capsid is certainly an evolutionary benefit, but the disease known as "scrapie" astonishingly seems to be able to survive without one for long durations.

We are quite lucky that nothing like that targets humans so effectively.

Yeah, because scrapie doesn't resemble a virus, so it doesn't have the same survival concerns. A protein in an exceptionally low-energy conformation is more robust than a ring of RNA, film at 11.

Anyway, fun fact, the prion that causes scrapie is extremely similar to the ones that cause mad cow disease (transmissible to humans) and Creutzfeldt-Jakob disease (human disease). So yes, we're very, very lucky, so far. Sleep tight.

Actually, the worst cases were from Kuru.

It has a ghastly origin.

https://en.m.wikipedia.org/wiki/Kuru_(disease)

(I will upvote you now.)

Sure, but Kuru doesn't transmit anywhere near as easily as scrapie does in sheep, which seemed to be the relevant comparison.

>That's like saying humans use gasoline combustion for our biology because we drive cars instead of walking sometimes.

I imagine you are speaking more in terms of molecular biology, but if we zoom out into evolutionary biology this is not such an unreasonable statement if we are talking in terms of the extended phenotype.

Especially as we get closer to a robust hypothesis that humans lack free will on a biochemical level, it really could be that our genes making use of gasoline, because the "we" that we like to think of as making use of it through intellect and choice is more a hallucination (perhaps useful in some way, perhaps mere side effect) than a biological reality.

Humans using petroleum isn't necessarily all that different from how a parasite modifies host behavior as an essential part of its lifecycle, or how a beaver builds a dam to have a much more survivable environment.

I like to think of a virus like code. It’s nothing more than a series of instructions in a file that, when executed (by the cell), does something. Is code complex enough to be called life?

We call self replicating programs viruses when they're unpleasant and quine as curiosities.

Is the game of life, life?

Something is getting lost in translation here, because scientists are not literally "debating" whether viruses are alive.

Scientists "discuss" the boundary between alive and not, which is an interesting question because it doesn't have a super clear and obvious answer. It becomes more a question of semantics and ontology than anything else.

What's really happening is that the words "alive" and "not alive" have outlived their usefulness: we need to think about things differently, and use better words, to fully capture how the world works. That's the discussion, and yes it's similar to what's happening with words like intelligence and consciousness in relation to AI.

A virus usually encodes for, and definitely utilizes, many processes of living things.

Part of the breakdown here is recognizing independent life requiring only a nutrient and energy source, and various consortia (e.g. can you say that a mitochondrion is alive without the host cell, and vice versa).

Basically, as mentioned by another commenter, the fact that "life" started out as a Platonic ideal that's being sussed out.

I'd say just change it to an adjective entirely and don't bother with a categorical noun.

"nutrient and energy source" is an interesting perspective: we certainly can't live off radiation (or heat, or electricity, or whatever one might consider as direct energy input) plus all the atoms contained in a lifetime worth of food arranged in simple monogamous molecules. We are no less dependent on other organisms arranging atoms into molecules our bodies can use as fuel and building blocks than a virus is dependent on a host cell for reproduction. I suppose that we can't really want a definition of life that excludes everything that isn't a plant?

As you state, the carbohydrates, fats, and proteins in our food sources function as energy sources. Photons are great for plants, but comparatively high-energy chemical bonds are fine, too.

> nothing could end that debate

Oh yeah, it could. If AI is textbook-defined technically “sentient” then that is perfectly enough to put an end to unethical abuses of sentient beings. (Imagine what that will do to the industry.)

It doesn’t matter if people still debate it. People still debate race supremacism, slavery, etc., that doesn’t mean a single thing as far as law is concerned.

Reply on news.ycombinator.com