roxgib

I have a Stream Deck XL connected to the TS3, seems to work fine (although I'm using BetterTouchTool to power it rather than the default software)

I got one of these recently, it's great. If you don't have too many peripherals you can achieve the same thing with a USB-C hub, but the TB ones obviously have the bandwidth you need for multiple monitors etc.

If someone is overweight, but otherwise healthy (which I suspect is the case for a lot of young people, as they haven't been overweight long enough for it to take a big toll on the body) can they not go ahead and enrol them and then whip them into shape during training? If they are really aren't able to meet their recruiting targets it seems like a missed opportunity.

I have a $4/month VPS that comes with a static IP address. Any reason you shouldn't use that as a proxy to solve the dynamic IP problem?

I've noticed the same. Never seen it get overloaded before, but it's happened several times over the past week.

Is that what we want? It's usually practice to have the unit price and amount on any invoice or receipt, this isn't something they invented to trick you. Honestly it's easier for me to reason with unit prices than totals.

In the author's defence, they do start the article by calling themselves an idiot.

Thanks for doing the research!

For deep learning/NNs, Google Colab/Kaggle notebooks are where most people start. There are lots of cloud providers that are harder to set up on but might be cheaper/faster for a particular project, but while Colab is limiting in some ways it's dead simple to use so you can focus on learning data science.

There's a lot of talk about large models atm, but when you're learning you should probably stick to smaller models because the training time drastically reduces how fast you can iterate, and most jobs aren't NLP based anyway.

Remember also that 99% of problems classical models are faster and just as good or even better than neural nets. Of course if you're interested in neural nets as a hobby, go for it, they're truly fascinating and you can do some amazing things just in Colab, but they still aren't the best option for most real world problems.

The obvious response to this is that the non-techy family members wouldn't be able to use it, but honestly most home entertainment setups are already crazy complicated, whereas everyone knows how to use a PC (for now)

I dodged a somewhat similar experience with Zoom. Took a contract with a company, and they tried to adopt my personal zoom account into their organisation (I was a contractor, so I was using my own email address while working for them). Fortunately I realised this might be annoying down the road and sent them an alias to use instead, but I imagine other people could get into strife - not that a Zoom account is a big deal and they seem to have better processes in place so it's possible it's easy to resolve, but it could easily cause a headache for a less tech savvy user.

In general, be careful what you click agree to (I know, I know)

I'd say this would be capable of fooling even people who are knowledgeable about USB attacks, because they're likely to recognise that it's a YubiKey presume it's not dangerous. Of course they'd also have less reason to plug it in.

> Another interesting gail.com factoid: my amazing e-mail provider, ProtonMail, rejects about 1.2 million mis-addressed e-mails per week to the gail.com domain.

Bloody hell, there could be some spicy stuff on a few of those emails!

I would hope that if domain matches your legal name, that should be sufficient evidence that you registered it in good faith. Anyway, I don't think the above poster was suggesting advertising it for sale, just that if someone reached out wanting to buy it that OP should seriously consider the offer.

To further clarify:

www.bom.gov.au is HTTP only. It will fail to connect if you use HTTPS.

reg.bom.gov.au is an HTTPS version of www.bom.gov.au. It seems to be staying, but you have to know to use that address.

weather.bom.gov.au is a newer site with a much better design, and supports HTTPS. It is being retired.

Obviously there are a million third party weather services/apps but they all use the BOM's data, at least in Australia.

The S is not a typo. It's retiring its HTTPS site. The old HTTP site (and the app) is staying.

You can actually get the old site via HTTPS, but only on the https://reg.bom.gov.au subdomain. So there's that.

Wouldn't have helped him here, he was 50 miles from the trail he wanted to hike!

He probably believed he was halfway through or more, but he was on the wrong trail.

I think he walked a lot further than that. The article said he had a 14 mile walk back to the car, and that he was a few miles into that when his legs started to cramp.

I think the question the court was trying to answer was "did he act negligently?". If he did, he's responsible for the result. That he didn't bring a first aid kit is further evidence that he wasn't prepared for the hike, so I think it's relevant.

For a while during covid NSW was closed, and a few people who had taken a road trip from Victoria to Queensland tried to drive home through NT and SA. The authorities spent a while rescuing people who weren't prepared for 100s of kilometres of dirt road between stops.

> They generally do not charge people unless they have substantial, if not overwhelming evidence.

Well that's simply untrue. Very often prosecutors will overcharge to force a plea deal on the defendant who doesn't want to risk it, and overwhelmed public defenders don't have the resources to properly examine the evidence or collect their own evidence in defence of their clients. And often it's a fraction of the sentence, not 75%, because charges are often dropped as part of the deal.

It seems to me that a few things are needed:

1. Limiting the 'trial penalty' to 25% or so of the sentence 2. Banning pleas that involve dropping charges in exchange for pleading to others 3. Proper funding for public defenders

The first and second are to solve the problem that prosecutors overcharge and defence lawyers advise innocent clients to plead guilty rather than risk significant jail time fight bogus charges. This seems to be a common tactic by prosecutors in the US and it should simply be banned through these mechanisms.

Of course the other problem is that people can't defend themselves against bogus charges because the cost is ruinous, so proper funding for public defenders is important.

This makes sense for companies as they tend to want to avoid bad hires more than they want to hire the 'best' person. As another person pointed out though it really sucks for people who happen not to have contacts, and often makes it harder for minorities.

Ideally it would be smart about the skills, and recognise how skills transfer between technology. I feel like this is a solvable problem with the right technology.

Your second point is important, but when you have recruiters/HR as the front line of hiring it completely breaks down because they haven't got a clue whether an applicant's skills are transferable or not. And they cop a lot of flack for this from developers but honestly I'm not sure it's possible for someone who doesn't even code for a living to learn how different technologies are and aren't compatible, skills-wise. Same reason we end up with unhelpful job descriptions. I think most developers would already be onboard with your second point but it's simply not up to them.

Many (most?) projects bypass this by simply not specifying which platforms are supported, and making decisions about breaking support on an ad hoc basis. Of course OSS developers are entitled to do this if they think it's best, but that's a good way to either create a burden for yourself when people start using your package in unintended ways, or to annoy people when something suddenly stops working because you never intended to support it.

Rust is doing it right by explicitly stating which targets are supported and to what extent. You can use it for other targets (I've written a lot of code for AVR which isn't supported at all) but if something breaks they stated up front that it wasn't supported.

I'm very confused by how this occurred. Did the school implement a policy that it's okay to send out AI generated emails so long as the use of AI is disclosed? Did the sender not proofread the email and notice that it mentioned ChatGPT at the bottom? What other emails are being sent out without a human so much as bothering to look at them?

A reminder that the law, and the bureaucracy, don't keep up with the latest social norms. This kind of work might be slowly becoming more socially acceptable, but it's easy for people who surround themselves with others who approve to forget that most of the world, even in liberal countries, still views it very negatively. Not judging, just worth pointing out. I sometimes see the same thing with drugs - people are surprised when their drug use causes them issues, because they think everyone just accepts in now.

Reminds me of this story:

https://www.nbcnews.com/tech/tech-news/robocop-park-fight-ho...

Turns out the 'police' robot isn't actually able to summon police when requested. It's literally security theatre.

Presumably one of the benefits of running your own model is that you can feed extra data into it via training rather than purely through inference? I.e. if you're a software company you could fine-tune it on your codebase, improving its answers without increasing inference time?