dwallin

I do want to point out, gaussian splats don't really offer anything in particular for realtime relighting, if anything it adds additional challenges. Under the hood, most implementations leverage spherical harmonics for baked-in lighting.

Did you by chance mean anisotropic spherical gaussians? ASG is a new-ish technique often used to model specular lighting but is unrelated to gaussian splatting.

From an amazon employee reply in the thread:

"When we say “Manufacturing cost”, it means your cost to source a product from a manufacturer/wholesaler/reseller, or produce the item if you are the manufacturer. You can provide the proof of your cost of sourcing and we will reimburse you accordingly. If you do not wish to provide your cost, we will provide our cost estimate and we will reimburse you for it. We calculate our estimate by evaluating the sourcing cost of comparable products sold by Amazon, by other sellers, and through other wholesale channels.

It excludes costs such as shipping, handling, customs duties, or other costs."

You could possibly try setting up your own "reseller" entity, where they act as an in-between, handling the shipping costs, etc and then reselling it to upon arrival for near the price sold?

I didn't say the consequences should be proportional, I said the responsibility (or blame); and profits derived should just be one part of the equation.

Criminal consequences of a company collectively being responsible for the deaths of individual(s) shouldn't just disappear because no one individual ostensibly caused it. A company is a system of incentives and processes, and if those incentives and processes are leading to preventable deaths and suffering, those who benefit most from and are most responsible for perpetuating those incentives and processes should be held liable.

I would seriously wonder how much you understand about proportions given the average nurse makes about 00.000002% of the $4.5 trillion in annual health care spending in the USA.

I also didn't say that should be the only factor. Clearly proximate responsibility for the relevant decisions should be (and already is) part of the equation.

Responsibility for the actions of a company should be somewhat proportional to the amount of profit derived from said activities.

All signs point to this being fake.

All evidence points towards this being post being fake.

Trying to reduce a negative by pumping resources into a positive rarely works out as expected and often has surprisingly distortionary effects. (see ethanol and corn production in the USA)

I’m personally of the opinion we should be doing far more tying together of revenue neutral taxes and subsidies within an industry. When you want to reduce a negative externality you tax that and then redistribute the proceeds equitably back across relevant actors. When you want to increase a positive externality, you equally tax actors and then distribute it asymmetrically according to the behavior you want to encourage. Or combine the two approaches to address both negative and positive externalities in one go.

These approaches allow you to be more targeted, while minimizing overall market distortions.

Yeah, this is one of the big issues with digital peer trust networks. Trust in human interactions is fuzzy and very conditional, which is hard to collect, represent, and update in a user friendly and low effort way. Hence we tend to collapse it to a single, often-binary, trust score.

They specifically mention a coating in the abstract, parylene C.

You could just as easily say “people don’t matter, people are just a collection of cells. You can’t predict a person without understand how their cells behave”, which clearly is a ridiculous statement. Structure really matters, at all scales, and the way a collection of entities behaves often diverges significantly from what you would get looking at individual entities in isolation.

This almost reads as satire, but I assume it’s low quality ai or just auto-generated content sourced from an FDA recall.

But it just simply is an argument against immigration, which we can totally have but which Is a different discussion. Unless you are suggesting that you want increased immigration, but for them not to be employed, which I think would be an extremely niche position.

But the data evidence we are discussing doesn't show any squaring of your statement at all. The only thing it shows is that the the amount of foreign-born individuals has increased, and that they hold down jobs at essentially the same rate as native-born Americans. If the data didn't look like that people would be complaining that immigrants are lazy, and don't work and are dragging down the economy. It's all a big nothingburger.

Relightable Gaussian Codec Avatars are very, very far from your off-the-shelf splatting tech. It's fair to say that this paper is more about a way of generating more efficiently, but in the original paper from the codec avatars team (https://arxiv.org/pdf/2312.03704) they required a A100 to run at just above 60fps at 1024x1024.

Nothing here seems to have moved that needle.

Here's what you are missing, the relative populations of native born and foreign born have changed over that time period. If you do the same thing for the population data (https://fred.stlouisfed.org/seriesBeta/LNU00073413) you see that the graphs are almost exactly the same. % of foreign born population goes from 17.5% to 22%, versus the 18.3% to 23.7% you called out for job change.

Given the complete lack of any actual details about performance I would hazard a guess that this approach is likely barely realtime, requiring top hardware, and/or delivering an unimpressive fps. I would love to get more details though.

Your wife's graph is massively misleading. Why would you choose to put two different scales on the y-axis when they are already in the same units? The reality of the data you linked to is that the 5 million job difference you claim is pretty much an arbitrary artifact based on whatever month you place your starting line, the amount of native jobs is essentially flat from pre-pandemic. The amount the foreign-born jobs changed is on the same order of magnitude as seasonal fluctuations in native-born jobs and would barely register as a blip if you used a fair and consistent scale.

That's clearly a false dichotomy, even the post you are replying to outlined a number of ways which an artist might supplement their income.

Also the idea that a significant UBI would lead to people just coasting seems patently false. Every bit of evidence shows that when given the opportunity the majority of individuals are driven to improve the quality of their lives; and also, that quality is measured in relative and not absolute terms.

5-6% of sales ABSOLUTELY matters. The biggest question is really will the majority of these purchasers substitute the absent "Mini" model with another iPhone model? If 90% of those users will just buy a different model then there is little advantage to offering the mini, given the additional design/engineering costs of the smaller size.

Anecdotally, I'm a techie but a small phone die-hard. I help onto my original SE until they finally released a mini model and will likely do something similar with mini 13. They definitely lost some potential upgrade sales from me, but despite that Apple STILL has the best tiny phone offerings, they are still software and hardware supported by Apple, although you have to purchase refurbished. There just isn't a serious Android competitor, with any sort of long term commitment to small phones, that would justify the switching cost.

My opinion is that Apple botched their positioning with the mini. "Mini" is a diminutive term, it implies less than. It was the lowest pricing outside of their SE model. They released the 12 mini shortly after an SE release, leading to cannibalized sales. And they released it at the beginning of a period (ongoing) where Apple leaned into the messaging of "Bigger is Premium". The only people left to buy the mini were people obsessed with small phones.

I think a more successful product would have been to lean into their "Air" brand, leveraging Apple's known ability to engineer premium devices in the smallest, most-elegant packages; positioning it as a premium product orthogonal to their main iPhone line. And from the rumors, it sounds like this might be Apple's next move in a device market that is increasingly undifferentiated.

Regenerative braking is so far from being an innovation on Tesla's part. Regenerative braking as a technology has been around since 1886, and was even seen in some of the earliest electric cars at the end of the 19th century (krieger electric landaulet).

Even when you fast forward to modern production cars, the Prius had a regenerative braking system almost a decade before the Tesla Roadster came out.

I'm confident they could come up with a filler eye animation algorithm that was convincing enough to pass muster for short periods of time. Even if hand coding something didn't quite work out, they certainly have tons of eye tracking data internally they could use to train a small model, or optimize parameters.

It's notable that you completely skipped over the following point:

> Two-thirds of the members fired by New York Times management since the Times Tech Guild formed have been from underrepresented groups.

The link to the actual decision: https://cases.justia.com/federal/appellate-courts/ca3/22-306...

Is it just me, or is this bad statistics?

With a small sample size and large numbers of personas / categories you would expect to see a positive bump, even if there was no statistical relationship between the persona and the preference. Since you are only eliminating categories that don't happen to be represented in the subset you are testing, you can only ever actually go up.

For demonstration I rolled 20 dice randomly for 6 personas and 3 categories of preference:

1, 5, 4, 4, 4, 2, 6, 1, 4, 2, 1, 2, 5, 6, 3, 6, 3, 2, 1, 3

A, A, A, A, B, C, B, A, C, A, C, A, B, C, A, A, B, A, B, B

A = 10, B = 6, C = 4; Which gives me 20% for C

I restrict myself to just the numbers that voted for C (2, 4, 1, 6) removing all 3 and 5s

I now am left with:

1, 4, 4, 4, 2, 6, 1, 4, 2, 1, 2, 6, 6, 2, 1

A, A, A, B, C, B, A, C, A, C, A, C, A, A, B

This now gives me A = 8, B = 3, C = 4

And now I get 27%, a nice 35% boost! Even better than Superhuman's 10% boost. But this is all an illusion, there was absolutely no dependency between the persona and preference here, which you would only see with a large enough sample size.

Searle's argument in the Chinese Room is horribly flawed. It treats the algorithm and the machine it runs on as the same thing. Just because a human brain embeds the algorithm within the hardware doesn't mean they are interchangeable.

In the Chinese Room, the human is operating as computing hardware (and just a subset of it, the room itself is substantial part of the machine). The algorithm being run is itself is the source of any understanding. The human not internalizing the algorithm is entirely unrelated. The human contains a bunch of unrelated machinery that was not being utilized by the room algorithm. They are not a superset of the original algorithm and not even a proper subset.

Actually an additional point, If you read the report, the 80% failure rate claim does not come directly from this report, but links to the following article as it's source: https://fortune.com/2022/07/26/a-i-success-business-sense-ai... (https://web.archive.org/web/20240105144440/https://fortune.c...) Which actually further quotes ostensible sources without linking: "That’s borne out in a slew of recent surveys, where business leaders have put the failure rate of A.I. projects at between 83% and 92%."

Unless someone is able to track down the original source of this information I would treat it with great scepticism.

Most of those non-AI projects are likely using well-established practices and technologies. From that perspective a 20% success rate seems pretty good to me.

An interesting note:

> By some estimates, more than 80 percent of AI projects fail — twice the rate of failure for information technology projects that do not involve AI.

So 60% general success rate vs 20% for an emerging technology that doesn't really have established best practices yet? That seems pretty good to me.

It's probably better to just link to the Rand Report: https://www.rand.org/pubs/research_reports/RRA2680-1.html