r00ty

On the one hand, yes. But on the other hand when a price hits a low there will (because it’s a prerequisite for the low to happen) be people selling market to the bottom. On a high there will be people buying market to the top. And they’ll be doing it in big numbers as well as small.

Yes, most of the movements are caused by algorithms, no doubt. But as the price moves you’ll find buyer and seller matches right up to hitting the extremes.

AI done well could in theory both learn how to capitalise on these extremes by making smart trades faster, but also know how to trick algorithms and bait humans with their trades. That is, acting like a human with knowledge of the entire history to pattern match and acting in microseconds.

Yep. Maybe they want to keep the MAU figure up while getting all the data from the majority that use normal / the app.

I know I only use old and not logged in if there’s something I need to see on there.

I love that this is the solution to all the logged-out user crap they’ve pulled lately.

Alas, I suspect it means they will drop old. in the near future.

In certain devices (batteries and power supplies) there’s a minimum weight that can realistically store or convert a specified amount of energy or power.

So if you buy a 1000w PSU and it’s too light, you’re going to know it is fake. So they add the weight to make it feel right for the power rating. In this case this is a double-whammy of a failure waiting to happen. A PSU with a lower than advertised rating, coupled with a lack of safety circuitry means it’s more likely to fail due to the overload applied, and when it fails it’s more likely to go out in a big way.

I think the point is. If you buy a cheap GPU it’ll either be a fake (lower spec with borked firmware) or lower spec branded. So the worst that happens is you have lower FPS, or it just doesn’t work. Same with all other components. They’re rarely off spec to the extent they will damage other components.

But a cheap switched mode PSU? Yeah the failure mode of switched mode supplies without proper protections is a high voltage on the rails feeding your components. They can take out your board, GPU, Drives and depending on what protections the mainboard has, the CPU and RAM too. Not to mention your precious RGB!

I remember back in the 90s/2000s we had a “server” where I worked at the time. I say “server”, the company cheaped out and had a high street PC builder make them. They were essentially desktops in a bigger box with expensive CPUs and things like tape drives. But yes, they cheaped out on the PSU and it popped. It took out a £1k Tape drive, about the same value in hard disks, and pretty much everything else that was connected.

It was not cheap to get that back up and running, I can tell you.

Sounds a bit too close to a black mirror episode to me. The one with the dating site.

Yeah, if they can, they will.

Aside from me obviously joking about home 22Pb/s. I’ll tell you a story.

In the early 90s. I was gifted a modem. I used it to connect to BBS systems. It had two outgoing call speeds. 300bit/s sync and 1200/75bps. And since I’d never connected my computer to the outside world, the whole thing seemed amazing. Many BBS’ had modems that couldn’t cope with 1200/75, it was an odd speed. So, I had to swap to 300 for those. But it was still an amazing time. But, file sizes were pretty small. It really wasn’t a huge problem.

In the mid 1990s, when I had a 28.8k modem (and later 56k) to connect to the internet (and by the way, paying for the phone calls too). The always on 64k leased line (£1k per month) seemed like a dream. The 10Mbit coax ethernet at the office seemed like light speed. Hell, it was faster than the local hard disk (caching improved that, mind you). I remember in the office where we had an ISDN modem setup, chaining them together to get 128Kbit seemed like light speed. Later we got a 256Kbit leased line in the office, and it was amazing in terms of speed. I actually ran effectively a mini ISP that some of us connected to, to get free internet. It had 4 USR Courier modems.

Then, in 1999 I was lucky enough to be enrolled on a trial. A trial for 2Mbit ADSL. It was amazing, 2Mbit down, 256Kbit up. This was groundbreaking. Web pages loaded instantly. Everything was so much faster. LANs in the office and at home was up to 100Mbit, and that seemed pretty damn fast. We’d be sure we’d not need more than that.

Then, 8Mbit DSL arrived. Again, amazing leap forward. Gigabit LAN became the norm, and again, who would have thought it would be too slow for anything? After all, we were all using hard disks, and they really weren’t that fast after all. ADSL2+ arrived with speeds up to 24Mbit, and those of us unable to get the higher sync rates started to suffer the internet being just a bit too slow.

Then we got VDSL and faster cable internet. 80Mbit, 100Mbit, 150Mbit. These seemed overkill for many. But pretty soon we were downloading 4k video and 150GB games onto 5GB/s+ SSD drives. This started to feel slow to some.

Now we’re at a place where FTTH 1Gbit is becoming quite common. Many ISPs here in the UK are offering packages with speeds between 2 to 5Gbit/s too. The tech they use is apparently good for up to 50Gbit/s.

Now, with this history of speed increases leading to demand increases. Why do you think it will stop at 1Tb/s? Maybe we cannot currently imagine why we’d need such speeds. But, someone will find a way to fill such connections. Don’t limit yourself to just expanding what we do now.

Maybe you’re right, but I honestly would never say never when it comes to computing.

22Pb/s FTTH when?

It’s a real shame that IPv6 multicast is as IPv4’s multicast implementation (global multicast not propagated by ISPs). I feel like live streaming/live TV over the internet could have made really good use of this feature.

It would in theory mean that they wouldn’t even need to operate their own servers in other countries/regions. One server could feed everyone.

The problem is a real one, Netflix for example have ISP level caches so that their content isn’t streamed across the open internet when it doesn’t need to be. But in the rest of the world it’s been resolved differently. I’m honestly not sure how ISPs keep up with the backhaul requirements for an ever increasing requirement for speed.

Here’s what I’m going to say here. With Windows it’s very easy to make it a very slow running/system with problems. But, it’s generally quite hard to entirely break it such that you cannot get to the GUI and attempt to fix it.

With Linux, just updating will sometimes break the system to the extent that if you’re lucky it will boot to a terminal. I’m experienced with linux (since the 1990s) and I’ve had linux systems that took my a better part of a day to fix. Someone that just wants to turn it on and work is going to be lost trying to fix this kind of thing.

Ubuntu upgrades from one release to another are extremely hit and miss in my experience and again if you don’t know how to pick up a failed upgrade and complete it, then fix the broken dependencies, fix the upgraded stuff that doesn’t like your old config files, etc etc. You’re going to be in trouble.

Linux is objectively better in every way except when it goes wrong. This is one of the reasons normal users won’t adopt it en-masse.

So, I’m not sure this is new. Unless this article is several years old.

These have been around for a fair few years, and it’s a pretty cool idea. Big Clive did a teardown video of a set 2.5 years ago.

So, unless this guy invented this thing considerably before that, I’m going to say it’s a tenuous claim at best.

On the one hand, I don’t really know enough about AI to comment. What I do remember is that, Bill Gates said the Internet was just a fad in the 90s. This comment caused myself and others problems promoting the Internet in workplaces because those in charge for some reason put some weight to his words. :p

I think he was always like this. It’s just he’s now ALWAYS in the limelight, and it’s much more visible.

Yeah, I think I’ve lost track how many articles and youtube videos about amazing “solid state” batteries are just around the corner. But I’ve not seen one actually materialise.

I mean, it’s great if true. But, I’m going to wait and see.

There’s been a steady increase in lithium based technologies though. But I do wonder when and where the plateau there might be.

I think the problem, is with how fast this is moving in regard to both software and hardware and how accessible something that could easily be weaponized to normal people.

That’s the potential for trouble I can see, for example Auto-GPT with improvements on the tool itself and its dependencies could make for something pretty powerful that could be set loose with a very wide remit and absolutely no limit to what it will do, to achieve its goal.

Time will tell of course. But this is probably one of, if not the fastest moving topic in tech we’ve ever had.

I think enforcing complex characters is outdated. Allowing them is enough, since someone brute forcing still needs to consider them. Of course they could try all lower, then mixed, then including complex characters in that order to catch those that don’t. But still, it’s better to have a password made up of compound words that is longer, than S0meth!ngV3ryC0nvolu73D. Or just pure random (aka password generator)

My main issue is places that have a maximum password length. This is firstly a limitation on security, but more importantly throws a red flag because of the potential reasons for having a password length limit!

Nope. I think you’re not really understanding what I’m trying to say. I’m saying that ethics do not factor into an organisation’s decisions in the same way it doesn’t for a colony of insects. They are ethically neutral in that respect.

At the same time, if you apply ethics looking from the outside in, of course you will cast their actions as ethical and unethical and many of their actions will be unethical.

I’m actually saying this is a bad thing, but is just a property of how an organisation, and especially successful businesses, operate. We’re not going to change that, I suspect. As such we should expect businesses to exploit AI to the fullest ability, even knowing that removing most or all of their employees is bad for the employee, bad for the country (and the world), bad for the economy and ultimately in the future, bad for the business/organisation too. But they simply do not look that far ahead.

My point is, you don’t see insects as ethical or unethical. I see organisations the same way. They’re acting on instinct, and are just aiming to do what they exist for. Make money. Ethics don’t even come into it. Now, peering outside in, you can try to cast society’s ethical views on organisations. But, they generally don’t even consider them (until they are forced to by local legislation, or that the route to making more money, or indeed not less money is to be seen to be ethical).

This is why there’s more often than not a certain kind of person drawn to leadership positions.

I don’t really see organisations as unethical. They usually don’t act ethically, but that’s not because as a whole they’re unethical.

I see them more like insects. They generally react to stimuli and just do the same as the other insects/organisations, things that have been proven to work. They’re also generally driven by one basic instinct, to make more money, and they do it at any cost. The drones (employees) are entirely disposable in this endeavour and if they can entirely remove them from the equation they will do it in a heartbeat.

Even those that perhaps do have some form of ethical streak and don’t think they should dump all their employees for AI/robots? Well, good for them, but they’ll be driven out of business by those that do.

When you think of a business or other organisation in this way, a lot of the weird things they do start to make a lot of sense.

I think ultimately this is going to become the crunch point. Because what kind of jobs can AI eventually take over (with appropriate robotics) in the mid-term future?

• Driving (if all cars were computer controlled today and roads were segregated from pedestrians, it’d probably already be possible)
• Likely end to end delivery could be automated. Large amounts of the process already are
• Train (and bus based on item 1) drivers. Currently, much of the urban transit systems around the world are ATO, where the train controller opens/closes doors and starts the train and is primarily present for safety. The rest is done automatically. There are already fully automated transits, and I suspect it is unions and legitimate safety concerns stopping full automation. But, it could be done with some work I think.
• Software development. I mean, currently the AI prediction in Visual Studio is sometimes scarily good. It DOES need to be guided by someone that can recognise when it gets it wrong. But so often development of a function now is writing 2 lines and auto completing half of the rest of the lines from the “AI”. It’s really a task of improving LLM and tying in LLM to product specific knowledge. Our days are most certainly numbered I think.
• Software design. This is similar to the above. With a good LLM (or General AI) loaded with good product knowledge, you might only need a few people to maintain/rework requirements into a format they can work with and feed-back mistakes until they get a sensible result. Each time reducing the likelihood that mistake will happen again. We’ll need less for sure.
• I think a lot of the more basic functions of a nurse might well become tasks for some form of robotic AI companion for fully trained nurses/doctors. Maybe this is a bit further away
• Airline pilots could probably already be replaced, and it’s purely on the safety grounds that I’m glad they’re not. Generally once a route is programmed the pilots on a flight that goes well, will drive the plane to the runway, the plane will automatically set thrust for economic take-off. Once established in the air autopilot will pretty much take them to their destination. Pilots can then switch modes, and the autopilot for an equipped airport can take the plane to a safe landing. Although in practice, pilots usually take control back around 500 feet from the ground, I think. It’s not really many steps that need automating. I feel like, at least one pilot will be retained for safety reasons. For the reasons for certain high profile incidents, there’s an argument to keep 2 forever. But, in terms of could they be replaced? Yes, totally.
• Salespersons. Honestly, the way algorithms trick people into buying things they don’t need. I’d argue they’ve already been replaced and businesses just still employ real sales people because they feel they need to :P
• Cleaners (domestic and street/commercial) could potentially be replaced by robotic versions. At the very least, the number of real people needed could be drastically reduced to supervisors of a robotic team.
• Retail workers. There’s already the automated McDonald’s isn’t there? I also think the fact commercial property in large cities is becoming less occupied is a sign that as a whole, we’re moving away from high-street retail and more online or specialist. As such, while we’ll always probably need some real people here, the numbers will be much lower.

Now, when it comes to industrial and farm work. There’s a LOT that is already semi-automated. One person can do the job with tech that might have taken 10 or more now. I can see this improving and if we ever pull of a more generalised AI approach, more entire roles could be eliminated.

My main point is, we’re already at the point where the number of jobs that need people are considerably less than they used to be, this trend will continue. We know we cannot trust the free market and business in general to be ethical about this. So we should expect a large surplus of people with no real chance of gainful employment.

How we deal with that is important. Do we keep capitalism and go with a UBI and allow people to pursue their passions to top that up? Do we have some kind of inverse lottery for the jobs that do need doing? Where people perhaps take a 3 month block of 3 day working weeks to fill some of the positions that are needed? I’m not sure. I suspect we’re going to go through at least a short period of “dark age” where the rich get MUCH richer, and everyone else gets screwed over before something is done about the problem.

Looks to me like Gates is looking ahead at this.

Sorry if that wall of text sounds pessimistic. Just one way I can see things going.