Infinity315

I disagree with the assertion that the relation is cyclical. In reality, all of these systems are highly interdependent. I'd model it as a weighted complete digraph.

And certain subsets of these various nodes have a greater outsized influence than their peers. For example, the intelligentsia within the people are usually far more impactful than say Joe Blow from Appalachia.

Are you not making the positive claim that LLMs can (eventually) generate good incident reports?

Please refer to this: https://en.wikipedia.org/wiki/Burden_of_proof_(philosophy)

Just to make sure we both understand what burden of proof is:

Suppose two people are having a debate over whether or not a teapot exists in the orbit of Jupiter which is impossible to observe via telescope. Where does the burden of proof lie?

Just to reiterate plainly:

Does the burden of proof lie on the person making empirically impossible to falsify claim or the person making the empirically possible to falsify claim?

Which of the following two claims is impossible to empirically falsify?

1. "LLMs can eventually be used to produce good incident reports."

2. "LLMs can never eventually be used to produce good incident reports."

Wrong. My original wager would be a successful deployment, but sure that could be interpreted as a weasel word.

What I mean by successful is that the LLM can generate accurate (>80%) incident response reports and propose correct fixes.

I'm fairly certain anyone literate could have read the rest of my comment and parse out what a successful deployment means.

> This particular hypothesis is obviously a testable one: someone could simply follow the proposed steps from the hypothesis (e.g. finetune a model on their incident response data), and see if it works.

Are you saying that if someone finetunes a current SOTA LLM with incident response data and demonstrates that it doesn't work that you'll say that LLMs are infeasible for this application? That would invalidate the hypothesis: "X application can be done on current LLMs."

Such a test could never invalidate the the hypothesis: "X application can (eventually) be done on LLMs."

If it's the former hypothesis you were asserting, then yes I agree that it is testable, but I'm fairly confident you were asserting the latter.

Earlier I had asked you: "I think we're at an epistemic impasse here. At what point would/could you be convinced that LLMs are incapable or unsuited here?"

And you have yet to provide a response.

It's a funny thing, because the inverse is falsifiable (testable) whereas the positive version is not. The inverse proof (I would say a hypothesis) is simply application of the scientific method.

There is a way to disprove that the statement: "There is no god" by simply showing a counterfactual god.

There is however no way to disprove the statement: "There is a god."

Likewise, there is a way to disprove the statement: "LLMs cannot be successfully used for X application." By showing that LLMs have been used in X application.

Again, there is no way to disprove the statement: "LLMs can (eventually) be used in X application."

The meat of my question was meant to demonstrate a failure to apply the scientific method.

It's not proof by inexistence, it's simply application of the scientific method--only the most successful method to date.

It seems like your assumptions are unfalsifiable. As computer scientists, I believe it's important that our hypotheses are testable. If a hypothesis is unfalsifiable, then the hypothesis is no better than theology and should be discarded.

What's stopping you and other VCs just pouring endless money into an idea that won't work?

Maybe. But in the list of lucrative applications I think bug-fixing is near the top. I think it's lucrative enough to attract at least a decent chunk of engineering talent.

I don't think it's necessarily self-incrimination to report a crime you witnessed, though I think it's dependent based on the time from when it occurred to the time of reporting.

You're technically right (I think). However, I believe if you witness a murder and know the murderer and the police asks you: "Do you know anything about X murder?" Then I think you're legally required to tell the truth here.

Because those services don't get shown reports of CSAM and then turn a blind eye to it and do nothing about it.

A person witnessing a crime by itself is not a crime. However, a person witnessing a crime and choosing not to report it is a crime.

Then it should be trivial to point out the error. So do it.

Alright, fine. Maybe you don't have faith in management, but perhaps you do have faith in the open market and capitalism.

Feel free to point out any error in my logic:

There are huge financial incentives--tens if not hundreds of billions of dollars--for developing an LLM which can solve novel bugs. So surely there exists AI companies developing an LLM capable of doing so. If an LLM capable of solving novel bugs exists, AI companies would rush to showing it off to capture tonnes of VC money. AI companies could show off their fancy bug-fixing LLM by closing issues on public Github repos using said LLMs.

No such mythical LLM exists. We are thus left with two choices:

1. My logic is flawed or there is an alternative possibility I haven't considered.

2. The LLM capable of doing what OP asserts doesn't exist and can't be made, despite their assertion that it is trivial to fine tune and put into application.

I think we're at an epistemic impasse here. At what point would/could you be convinced that LLMs are incapable or unsuited here? If LLMs were successfully deployed in a production environment is the day I bite my tongue. What about you?

I'm not even sure that LLMs are even capable of solving standard bugs see: [1]. Hallucination seems to be a significant hurdle and any time spent validating the fixes of an LLM is wasted when it could be spent tackling the bug head on. The amount of energy spent espousing garbage requires an order of magnitude more effort to invalidate.

[1]. https://daniel.haxx.se/blog/2024/01/02/the-i-in-llm-stands-f...

I take issue with this.

Typically issues arise because they are novel and are unforeseen. If we did see these issues beforehand they'd be fixed! LLMs by definition are trained by example, so I fail to see how finetuning LLMs on things that have already happened to be helpful for determining the root cause of a novel issue.

LLMs seem to lack systemic modelling that humans do. I can see LLMs being practical for this if it is shown that LLMs are capable of modelling scenarios outside of their dataset, but thus far none such examples exist.

> Your logic would presumably interpret this as even naturals.. but ya

My logic comes from a Math background. Take any Abstract Algebra class and this notation is common.

For example: https://math.stackexchange.com/questions/1378174/how-to-deno....

To say the least I take numerous issues with the notation used, but I accept it. Since the integers and the naturals have the same cardinality, we have that Int = Nat.

> Nat has countably many inhabitants. And one more than countably infinite is countably infinite!

Two sets having the same cardinality (number of inhabitants) is insufficient to prove that two sets are equal. There is an additional condition that they all share the same elements.

I read 1 + Nat as: map each element in the Naturals to its successor (i.e. Nat - {0}). 0 is in Nat, but 0 does not exist in "1 + Nat" since it got mapped to 1.

This would be true if the set we were working with were the Integers instead of the Naturals.

It doesn't, but consider that the vast majority of us already carry our phones everywhere.

Would carrying an extra USB stick be that big of a hassle? No, but I can see the need for booting up a ready Linux image being extremely situational so the vast majority of time you're just carrying dead weight.

Is every word coming out of my mouth just incomprehensible? I feel like I've been really specific. How can you possibly interpret anything I've said as me believing every word out of her mouth? ** I am strictly going off what the EPA says ***

At the risk of sounding snarky, but it seems like you have a recurring pattern of just not reading what I've typed. First you've misunderstood the carbon filter and now. It's like I say X and you've heard Y and Z.

> I didn't see anywhere that they weren't changing the filters enough.

I don't think you've read observation #11 or my comment.

From observation #11:

> Apple does not appear to have included all the solvent waste streams when calculating the breakthrough times for the "Activated Carbon" boxes to ensure no VOCs are released into the atmosphere onsite.

Breakthrough times are a function of how much VOCs you are putting through them. If you under account for the amount of VOCs you will overestimate the "true" breakthrough time.

With regards to 1 and 2: This is inconsequential to what the EPA reported. It doesn't matter whether they do R&D for plushies or whatever, fact is Apple was under accounting their waste stream.

> She says the carbon filters weren't changed, not the EPA.

I'm going by what the EPA says and not what she says.

> How is it this woman was nearly killed, while the employees who WORK in the building are perfectly healthy, the next door businesses have no sick employees, and the other thousands of residents in the apartment building are also not sick?

I don't know. It's a matter worth investigating. Also, it's perfectly consistent for people inside the building to be fine whilst working inside considering the toxic fumes are vented outside the building and aren't just circulating inside it.

> Why isn't the news covering this? Why aren't they shut down?

This is out of my wheelhouse. I'm guessing government agencies are bit more reticent with regards to shutting down businesses unless there is a readily apparent danger to public health (i.e. something that could hold up to scrutiny in court).

To be honest, I give very little if any credence to anything she's said, especially if it's unsupported by evidence. Notice I'm not relying on anything she's said. Even if she is mentally ill, it's not reasonable to always assume the opposite of everything she asserts; that's called being a reactionary.

Fact of the matter is, EPA reported that Apple was under accounting their waste streams leaving open the possibility that filters were not changed as often as they should have been.

> Probably because they don’t have the same needs that you do?

To be clear, my entire workflow could easily be done on my laptop. It's just that I found the downsides to be minimal to non-existent and there exists only upsides.

> My primary machine is a laptop and it’s very powerful without me having compromised on its portability to any meaningful degree. I enjoy the myriad benefits of not doing remote development, such as not requiring a stable internet connection at all times.

I think a suitable development environment almost always requires a stable internet connection. But I do live in a cushy first world country, so I admit I'm privileged in that regard. From my POV, this is the only potential upside to not doing remote development. I'm curious to hear other though!

I'm surprised more developers haven't adopted remote development. It's fairly simple to setup an SSH server on your computer and DDNS on many routers.

It comes with the added benefit that everything I need is accessible on any computer with a connection to the internet (AKA everywhere). Not to mention it saves power on your laptop!

Reposting here since the other thread got flagged.

Note, I am not a lawyer or anyone who could professionally give an opinion on this.*

The most egregious violation the author purports Apple did is the illegal venting of hazardous chemicals.

IMHO, the most plausible evidence supporting the author's claim can be found in the (redacted) EPA report in observation #11.[1]

My understanding is that waste streams (hazardous materials being vented into the environment) must be accounted for in order to accurately calculate how often you must change carbon filters (called breakthrough time in the report) and that carbon filters lose efficacy as hazardous gasses are passed through it.

Anyways, the report found that Apple did not account for all the waste they were putting through the filters and thus could not accurately assess the "breakthrough time" of the filters. There is a plausible case that could be made that a filter had reached its "breakthrough time" -- potentially leaking hazardous gases into the environment. This reportedly has been occurring since atleast* Dec 2020, which could be around the time the author moved in which was also 2020 (unspecified range). Note, this neither confirms nor denies the claim that hazardous chemicals did leech into the air.

1. https://www.dropbox.com/scl/fo/312oqvretg7yyhfx0ohx4/AGZduVX...

There is a disappointing amount of character slander rather than refutation of the author's claims happening on this thread. The foremost trustworthy party is the EPA.

Note, I am not a lawyer or anyone who could professionally give an opinion on this.*

The most egregious violation the author purports Apple did is the illegal venting of hazardous chemicals.

IMHO, the most plausible evidence supporting the author's claim can be found in the (redacted) EPA report in observation #11.[1]

My understanding is that waste streams (hazardous materials being vented into the environment) must be accounted for in order to accurately calculate how often you must change carbon filters (called breakthrough time in the report) and that carbon filters lose efficacy as hazardous gasses are passed through it.

Anyways, the report found that Apple did not account for all the waste they were putting through the filters and thus could not accurately assess the "breakthrough time" of the filters. There is a plausible case that could be made that a filter had reached its "breakthrough time" -- potentially leaking hazardous gases into the environment. This reportedly has been occurring since atleast* Dec 2020, which could be around the time the author moved in which was also 2020 (unspecified range).

1. https://www.dropbox.com/scl/fo/312oqvretg7yyhfx0ohx4/AGZduVX...

Slightly off topic, but this is demonstrated in a show called Mr. Robot. I find it insanely cool that the hacking in Mr. Robot closely resembles real life.

A method which this article doesn't mention is stochastic rounding or probabilistic rounding.

Rather than striving for exact values for every computation, we make it so that the expected value is exact. For example, suppose our number system was restricted to the integers and we have incoming value 1.1, with stochastic rounding we'd round down to 1 90% of the time and 2 10% of the time giving an expected value of 1.1!

Further reading:

https://nhigham.com/2020/07/07/what-is-stochastic-rounding/

> Nobody should be subjected to mandatory fluoride in their tap water.

Why? Just for the sake of freedom?

I think this raises an interesting question.

Suppose we have empirical irrefutable proof a practice is beneficial and doing so would not hinder anyone's lives whatsoever (as in, you don't have to change your lifestyle at all). Can this practice then be justified?

For example, suppose that fluoridating tap water (at least in the quantities present in most municipalities) was shown through many longitudinal studies to not have any adverse effects but was shown to have positive effects such as improved dental health. Would you support fluoridating water then?

> Payment infrastructure? All of it? Some of it?

The point at which the only way of transacting is through cash during the duration of this outage.

What odds would you give on something like payment infrastructure failing for 3 or more consecutive days happening in a year or 10 years?