pifm_guy

Although if apple didn't have Siri, then they would likely allow other voice assistants on the platform, and you'd have 'dictation by alexa' instead.

Here is a news report with video footage of one such incident (the part where the victim is decapitated is clipped from the report): https://youtu.be/3IwhsYfnNvs

I am personally aware of quite a few speedboat related accidents, some fatal, and this is a common pattern.

The types of death that people imagine (boat sinks, boat engine dies and drifts out to sea) tend not to be killer in small boats because they mostly operate in busy waterways near shore where someone else will come help.

False negatives are more worrying. On a small speedboat, a common cause of death is that the captain falls overboard while the boat is doing a sharp turn at speed. The boat then does a 360 circle and within 10 seconds runs over the captain before anyone else in the boat can intervene.

Will these wireless keyfobs reliably cut out within 10 seconds when the boat never goes more than say 60 feet from the captain? I suspect not.

You should also design all your code to be rollbackable... But for the very rare case that a rollback won't solve the problem (eg. An outage is caused by changes outside your organisation's control), you also need to be able to do a rapid code change, recompile and push. Many companies aren't able to do this for example their release process involves multiple days worth of interlocked manual steps.

Don't get yourself in that position.

Obviously you do everything possible to stop an outage like this happening...

But when it inevitably does, you should be prepared for a full system simultaneous restart. Ie. So that no 'bad' signals or data from the old system can impact the new.

That is the sort of thing you should practice in the staging environment from time to time, just for when it might be needed. It could have taken this outage from many hours down to just many minutes.

Twitter indexing was basically real-time - you could see a tweet within a few minutes of it being made

Mastodon seems to take many days to appear, if at all. I suspect Google just scrapes the webui occasionally and doesn't use any push-based mechanism for notifications. That probably means it misses some too.

I never really saw how making the counters random helped fight bots.

Anyone who wants to know if their bot armies upvotes are counted can just choose 2000 articles, upvote a random half, then see if the half they upvoted have higher vote tallies than the half they didn't.

No amount of delay, quantizing, or adding noise will defeat that tactic. So why try at all?

They didn't give permission for the internet service to launch there. But they did give overflight permission.

Remember that Loon, the USA Google balloon project, sent similar balloons around the world, including over China.

They didn't shoot our balloons down.

Ask someone local.

What's available and rules vary really widely city to city.

What makes sense financially also depends on your building construction, climate and fuel prices, which also depends a lot on location.

I just filtered for the word yeti.

I don't want a yeti hopper...

All of those things look pretty 'fixable' in the game of cat and mouse ...

I really want to see ssd manufacturers offer a decent warranty...

This drive costs $100, and will last 10 years or until 100TB has been written to it, as long as you keep it within the specified temperature/humidity/power conditions.

If it fails to do that, we will return $1000 to you.

It's a shame this is US only.

Typically companies that do that will later release in ~10 more countries, or maybe even 50 more, and then give up.

If you live in Mozambique, the chances of ever being able to buy this are pretty much nil.

Sketch of a design to solve this:

Company can upload some prompts to OpenAI, and be given 'prompt tokens'.

Then companies client side app can run a query with '<prompt_token>[user data]<other_prompt_token>'. They may have a delegated API key which has limits applied - for example, may only use this model, must always start with this prompt.

That really reduces the privacy worries of using all these third party companies.

It's the chicken and egg problem... If you started a new card provider today, you'd need to get consumers onboard, banks onboard, merchants onboard, and every card machine in the world replaced to accept your new card type.

The closest we get to that is loyalty cards and gift cards that only work in one shop. Some of those have managed to be compatible with a few shops, but nowhere near what would be needed for a consumer to consider it a replacement for visa.

PayPal does the same...

For non-us citizens, USD's either incur a conversation fee to another currency, or an international withdrawal fee. There is no way to get money out without paying one or other fee.

The patch causes a different root filesystem to be used. Specifically, if the kernel command line specifies a root filesystem that must be used, then that one will always be mounted, even if other devices are available.

could have involved the use of your phone number to send and receive phone calls

Surely from their logs they know if these calls/texts happened?

If, during that period no calls/sms's occurred, then there has been no breach - the attacker was close to their target, but walked away with nothing.

If messages/calls were made, the user really needs to know who they were to/from to make any informed decisions. And Google has those logs.

I think no company should be allowed limited liability/bankruptcy protection.

If a company can't pay its debts, the shareholders should be required to pay.

ChatGPT seems to do better on all these prompts...

All of these things are kinda a necessary consequence of the way the app works.

It's a social network... You should consider everything you do on it to be public...

This is a neat shower thought...

But I don't think it's worth writing an article until you have some results.

I'm fairly confident that something like this wouldn't have been discovered in the USA.

14,000 deaths by asthma like causes spread over 10 years in a population of 300 million is going to disappear into the noise.

There is nobody doing nationwide studies asking 'do people who have a humidifier live shorter lives than those who do not?'. Besides, any results from that question would be hidden behind spurious correlations (humidifiers are maybe owned by the rich, or used only in dryer places where disease profiles are different).

There are undoubtedly more things like this waiting to be discovered.

Killing large numbers of people 'statistically' is easy when nobody can identify a direct cause for any individual death. See: air pollutants.

I worry that data privacy laws will slow down the discovery of things like this.

Imagine for example that supermarkets could have said 'oh look, the people that buy this selection of products have a higher chance of hospital admission'.

Except now it's practically illegal for a supermarket (who has product data) do join that with medical records at any reasonable scale. And if you do a study of just a few thousand volunteers, you won't discover something like 'this floor cleaner brand shortens users lifespan by 1 year'.

Instead the problem goes undetected, and tens of thousands or millions die early.

FSD actually isn't a big part of what Tesla sells. They have $1B set aside from sales revenues of the FSD option to pay for implementation.

That sounds like a big number, but it's less than the revenue of 2 days car production.

If you buy anything online (in Europe at least), you can return it for a full refund within 14 days if you don't like it.

So why didn't they implement the HLT instruction as simply a 'jump to self' infinite loop?

Then no special logic would be needed, no extra states, etc.

Sure - there would be no power savings, and the memory bus wouldn't be idle, but we're either of those a requirement in 1970?

When I buy from AliExpress I feel like perhaps only 80% of items work properly for a reasonable lifespan...

I still buy there because I quite like the challenge of repairing the things that don't work, and prices there are typically half of what things cost on Amazon.

In Europe, by law it is the retailer responsible for warranty claims.

Many manufacturers will also offer warranty service, and it is up to you the customer which you want to use. Often manufacturers try to persuade you to contact them first because it costs them less than a refund would (typically retailers just refund rather than attempt a repair or replacement)