pwagland

The real question is how do you enforce that the human is reviewing and double-checking?

When the AI gets "good enough", and the review becomes largely rubber stamping, and 50% is pretty close to that, then you run the risk that a good percentage of the reviews are approved without real checks.

This is why nuclear operators and security scanning operators have regular "awareness checks". Is something like this also being done, and if so what is the failure rate of these checks?

In my experience, Gemini is great for "one-shot" work, and is my goto for "web" AI usage. Claude Code beats gemini-cli though. Gemini-cli isn't bad, but it's also not good.

I would love to try antigravity out some more, but last I don't think it is out of playground stage yet, and can't be used for anything remotely serious AFAIK.

And tear stains, or the lack thereof, are the only sure sign of quality.

Is there a fuller list of headsets that are affected being maintained anywhere? I could not find it. Since most manufacturers tend to reuse components, we can expect that more Sony stuff is affected, and probably more JBL/Jabra/Bose/Marshall that they didn't have access to.

Based on their timeline, full credit to Beyerdynamic!

Partial credit to Airoha, they took a long time to initiate the communications, but once they did, they seemed to take it seriously.

No credit to Sony and Marshall, as they either didn't, or effectively didn't, respond.

Unknown credit to Bose, JBL, Jabril, EarisMax, MoerLabs, and Teufel, as they don't appear in the timeline.

Go to the website, it has step by step guide for the supported robots: https://valetudo.cloud/pages/general/supported-robots.html

I tried this with Claude:

``` > [Image #1] Create a unicode "ascii-art" version of this image, with the optimal path through the maze highlighted in a solid colour.

I'll create an ASCII art version of this maze with the solution path highlighted!

  ┌─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┬─┐
  ●●│ │ │ │       │   │       │     │ │   │   │ │ │   │
  ├─┤●└─┴─┐ ├───┐ │ ╔═╗ ├─────┤ ╔═══╝ │ ╔═╝ ╔═╝ │ │ ╔═╝ ├─┤
  │ │●●●●●└─┤   │ │ ║ │ │     │ ║     │ ║   ║   │ │ ║   │ │
  │ └───┐●╔═╧═╗ │ └─╢ │ │ ┌─╔═╝ ║ ┌───┤ ╚═╗ ╚═╗ │ └─╢ ┌─┘ │
  │     │●║   ║ │   ║ │ │ │ ║   ║ │   │   ║   ║ │   ║ │   │
  ├───┐ │●╚═╗ ╚═╧═╗ ║ └─┤ │ ║ ╔═╝ └─╔═╧═╗ ╚═╗ ╚═╧═╗ ║ └─┬─┤
  │   │ │●●●└───┐ ║ ║   │ │ ║ ║     ║   ║   ║     ║ ║   │ │
  │ ╔═╝ └───┐●╔═╝ ║ ╚═╗ │ └─╢ ╚═╗ ╔═╝ ╔═╝ ┌─╢ ╔═══╝ ╚═╗ │ │
  │ ║       │●║   ║   ║ │   ║   ║ ║   ║   │ ║ ║       ║ │ │
  │ ╚═══╗ ╔═╝●╚═╗ ╚═╗ ║ ├─╔═╧═╗ ║ ║ ╔═╝ ╔═╝ ║ ║ ╔═╗ ╔═╝ │ │
  │     ║ ║●●●●●│   ║ ║ │ ║   ║ ║ ║ ║   ║   ║ ║ ║ ║ ║   │ │
  ├───╔═╝ ║●┌─┐●└─╗ ║ └─┤ ║ ╔═╝ ║ ╚═╝ ╔═╧═╗ ║ ╚═╝ ║ ╚═╗ │ │
  │   ║   ║●│ │●●●║ ║   │ ║ ║   ║     ║   ║ ║     ║   ║ │ │
  │ ╔═╝ ╔═╝●│ └─┐●╚═╧═╗ │ ║ ║ ╔═╧═══╗ ║ ╔═╝ ╚═══╗ ╚═╗ ║ └─┤
  │ ║   ║●●●│   │●●●●●║ │ ║ ║ ║     ║ ║ ║       ║   ║ ║   │
  │ ║ ╔═╝●┌─┴─╗ └───┐●╚═╝ ║ ║ ║ ╔═╗ ║ └─╢ ╔═══╗ ╚═╗ ╚═╝ ╔═╝
  │ ║ ║●●●│   ║     │●●●●●║ ║ ║ ║ ║ ║   ║ ║   ║   ║     ║
  │ ╚═╝●╔═╧═╗ ╚═╗ ╔═╧═══┐●╚═╝ ║ ║ ║ ╚═╗ ║ ║ ╔═╧═╗ ╚═══╗ ╚═╗
  │   ●●║   ║   ║ ║     │●●●●●║ ║ ║   ║ ║ ║ ║   ║     ║   ║
  ├─╔═┘●╚═╗ ╚═╗ ║ ║ ╔═╗ └───┐●╚═╝ ╚═╗ ║ └─╢ ║ ╔═╧═╗ ╔═╝ ╔═╝
  │ ║●●●●●║   ║ ║ ║ ║ ║     │●●●●●●●║ ║   ║ ║ ║   ║ ║   ║
  │ ║●┌─╗ ╚═╗ ║ └─╢ ║ ╚═══╗ └─────┐●╚═╧═╗ ║ ║ ║ ╔═╝ ║ ╔═╝ ╔
  │ ║●│ ║   ║ ║   ║ ║     ║       │●●●●●║ ║ ║ ║ ║   ║ ║   ║
  │ ╚═╝ ║ ╔═╝ ╚═╗ ║ ╚═╗ ╔═╧═════╗ └───┐●╚═╝ ║ ╚═╝ ╔═╝ ║ ╔═╝
  │     ║ ║     ║ ║   ║ ║       ║     │●●●●●║     ║   ║ ║
  └─────╨─╨─────╨─╨───╨─╨───────╨─────┴───●●╨─────╨───╨─╨──

  The path from the mouse () to the cheese () is marked with ● characters. This shows the optimal route through the maze, navigating through the corridors while avoiding all the walls!

```

Suffice to say, it didn't do either part right.

Cost is a big part of the reason.

A small-ish flan for 6 people costs more than than a "sponge" type cake for 12. And the sponge cake is massive in size comparison because of all the air.

And this is why Walmart, for example, will likely never sell proper bread, or other pastry products.

In most of Europe, single use bags are outlawed for white a long time already.

In my experience, to see how bad this can be, these days you need to go to Asia or Africa. Anecdotally, there is a lot less rubbish in Europe than there used to be.

1. https://www.greenqueen.com.hk/wp-content/uploads/2020/11/pla... 2. https://www.mirror.co.uk/news/world-news/gallery/horrifying-... 3. https://thebalisun.com/balis-most-popular-beaches-brace-for-...

Although latency is shockingly bad.

This is, but there is an almost identical file, assumedly for the non CN regions: https://github.com/BlueFalconHD/apple_generative_model_safet...

This is the same, except for one additional slur word.

This is a valid point missed by many today. The mantra of don't optimise early is often used as an excuse to not optimise at all, and so you end up with a lot of minor choices scattered throughout the code with all suck a tiny bit of performance out of the system. Fixing any of these is also considered to be worthless, as the improvement from any one change is miniscule. But added up, they become noticeable.

The author of the article freely admits that `std::function<>` is more flexible. He still prefers this solution, as it is easier for him to reason about. This is covered in the "Fringe Benefits" part of the document.

> You can't say an opinion is false - it's an opinion.

I agree that op thought: > [that] there's [no] debate that ATC staffing is a major issue right now, and the FAA really did blow up its hiring pipeline to further DEI goals

Grandparent only said that there is clearly debate over this.

To say that no-one can disagree with op, or point out an opposing point of view, because is their opinion, is not furthering discussion.

Given that http://susam.net:8000 has stopped responding, I suspect that there will be a lot more beeps today.

An interesting tidbit is that FSD is almost guaranteed to turn off before any accident occurs. Last I looked, the data was not available to see how long after FSD deactivation the accident occurred.

Note: I have an older Tesla, and actually quite like it. I don't have FSD, but I do have enhanced autopilot (EAP) and like it as well. That said, it is very easy to believe that people ignore the road for longer than they should with FSD/EAP turned on.

The grandparent comment also didn't mention the split between FSD miles/non-FSD miles. It is possible that FSD is so good, that every Tesla driver becomes useless when they are required to drive for themselves, and that is what drives the higher accident rate.

You have to insert 5c.

The free version of the IntelliJ IDEs are reasonably good, at least if you are using one of the languages that the free versions support.

The other advantage that the article misses for aliases is that completions will work through aliases. So, in the case give for `alias g=git` typing `g <tab>` will give you the bash/zsh completions for git. You _can_ do this for the script as well, but then you need to code that up separately.

But that really isn't true either. The model 3, for example, has undergone 3 or 4 refreshes that change a lot. They have not dramatically changed the exterior of the car, which I know a lot of people think is absolutely critical, since an 8 year old model 3 looks almost identical to a new model 3.

Most car makers haven't released new models in a long time, they just update their current models, and make them visibly different from the outside.

Now… if you were talking about the lack of new feature _in_ the car, then I would be happy to agree. And if you want to talk about the _still_ lacking functionality of automatic wipers…

>> How do you square that with the fact that spending as a percentage of GDP is only slightly elevated compared to the historical average...? > Why should it scale linearly with GDP? I can see an argument that it should scale linearly with population (maybe), but if GDP per capita increases, you could also expect better tech/productivity to allow gov spending per capita to decrease.

Over time it _can't_ scale linearly with population, unless you decide to not adjust for inflation. It _could_ scale with population and inflation, assuming that you agree that you don't want more services from your government.

Don't forget, a percentage of that GDP increase is just inflation.

Most people, as they get richer, want to have services increase, as they can afford to pay more. For example, they go to nicer restaurants, nicer hotels, maybe they get a massage, where previously they would not have, etc.

This is largely also true of a population. We expect that our children will be better educated. We expect better roads/bridges/other infrastructure. Heck, we might even expect better public infrastructure such as trains, buses, etc.

Sometimes, but not always. Sometimes it is also based on the locales in your browser.

Related to that, the first 100 days are typically when there is the highest number of orders per day. https://edition.cnn.com/interactive/2021/politics/biden-exec...

Tump did 20% of his total, from the first term, in the first 100 days. Biden did just under 1/3 of his total in the first 100 days.

Sadly, this is just another example of "A lie can travel halfway around the world before the truth puts on its shoes."

That doesn't mean that every sensational thing is a lie, but verifying the truth definitely takes time!

Well, that, and any organization that has gotten themselves into this situation tend to have a very strong risk aversion principal. Which means they _can't_ approve something like this organisationally since there is simply too much risk embedded, and someone has to accept that.

It would be convenient to reference the "why of a few days ago", since I, at least, was not aware of this from a few days ago, and so cannot tell you is the the same why, or an extension to the same why, or a different why.

No, it only makes it take longer to get the underlying secret.

Timing attacks are already dealing with "noisy" data, task scheduling et al, so they all boil down to some level of statistical analysis on the response times. Adding noise to that slows you down, but the underlying bias on the timings is still there.

Not quite, from the Wikipedia:

> Pegasus' iOS exploitation was identified in August 2016. Emirati human rights defender Ahmed Mansoor received a text message promising "secrets" about torture happening in prisons in the United Arab Emirates by following a link. Mansoor sent the link to Citizen Lab of the University of Toronto, which investigated, with the collaboration of Lookout, finding that if Mansoor had followed the link it would have jailbroken his phone and implanted the spyware into it, in a form of social engineering.

So the link was sent via text message, but you had to click on it. Receiving the text message did nothing in and of itself.

Because most people apply the software updates at some point, and this was fixed many years ago. Everything sold in the last years comes with a version of iOS that isn't vulnerable anymore.

The Hong Kong system's backend, and large parts of the frontend, was also written by an Australian company, ERG at the time, now VIX. So there was plenty of local skills that _could_ have been used.

Any source for this issue, I could not find any reference, but am not doubting that it exists.