Does that mean...
You're using a load of AIs to generate a list that you used to be able to get from a half-decent search engine?
1114 publicly visible posts • joined 16 Apr 2021
After the recent news, I decided to try it with some (more acceptable) queries.
As with the others I have tried, it generated a plausible but incorrect answer, and was more than happy to acknowledge it was in error when I told it so.
Summary - you can ask them anything, but you can't trust the answers and have to validate them yourself.
Yeah. I refused to install that and switched my business banking to Starling (which had the added advantage of restoring free banking*).
* It was the "pay a monthly fee for the privilege of being charged to send funds overseas with a pathetic exchange rate" that was the final nail in the coffin for them.
I do have some (very limited) sympathy for UK banks, as they are financially liable for fraud where an app is compromised, even if it wasn't their software that was at fault (for those not in the UK, that also includes where someone is tricked into transferring funds to a fraudster in response to a scammer calling them).
However, they should be assuming that the execution environment is potentially insecure, and making sure that their app does not make external (non-OS) calls or trust data that it cannot verify.
I use Pi_hole, and that totally breaks currys.co.uk. Filters do nothing, can't change the number of items that are displayed - only interaction that works are the menus and links.
Hmm. Looks like it's been updated (I reported the issue a few month ago) and it's now happy with alternate DNS / ad blocking.
Still, I have come across other sites that don't work.
Remind me - local hospital only allows Google to be used as a search engine. My phone is defaulted to duckduckgo, which gets blocked as "Threat category : Search Engines". Muppets.
From my reading of the article, it sounds like there are a number of software modules running and the cert is used to secure the communications between them (could be as simple as SSL). It makes sense to secure this, but I'm not sure a cert that expires is a good idea (it effectively means the device will be rendered inoperative some time ofter it "goes out of support"*).
* I do hope this is not a deliberate "feature" to force users to replace otherwise functional devices.
Anyone remember when "fuzzy logic" was the in-thing, and everything had to have it?
A lot of what is now called "AI" is really just some simple logic (that often isn't new).
What I would like is for Windows/Intel to run my code at full speed on a power core when it's trying to process data - specifically, I have a Python script that processes several million data samples where I have to set the power settings (manually) to "full speed" as it takes twice as long to run if Windows/Intel are allowed to manage the power state - a thread running at 100% on a CPU at the non-turbo speed when the rest of the cores are idle means the task will take longer than it should to complete.
I used to contract about 185 miles from home. Traffic on a Monday was not fun, so I used to set off very early and arrive at about 07:00.
I got in as usual one Monday, tried to login, and was greeted with a "You must change your password" dialog box that could only be dismissed either by cancelling or by changing the password. I obviously went with the second option, but all attempts to login after that were greeted with "Incorrect Password".
The hell desk didn't open until 08:30, so I happily spent the next (billable) 90 minutes drinking coffee and catching up on some reading.
At 08:30 I called the hell team, who asked if I had read the email they sent out the previous Friday instructing people to "cancel" the change notification as the login would still complete, and to then change the password from within Windows. Trouble with that was I was never in on a Friday and most people would have finished before the email was sent.
Their line was busy for quite a while that day.
I'm not a security expert.
Can anyone explain to me why it would ever, other than through laziness or stupidity, be necessary to expose an API endpoint like these to the internet? I guess "remote management" might be a use-case, but wouldn't that normally be provisioned over something like a VPN?
If there isn't a good reason, why is any feature that increases the attack surface even made available?
Yes, the ring was a solution to a copper shortage. However, as was stated above, the fuses in the plug protect the appliance's cable.
The fuse (or more likely these days, the breaker) supplying the ring protects the ring itself from overload.
Rings are normally rated at 32A - it doesn't take many moderate loads (say 3A) to overload a ring, which can easily have 10 double outlets on it (there is no restriction on the number of outlets).
However, there are further complications as unequal loading of a ring can theoretically lead to an overload if it is fully loaded at one point that is near to the origin - the load is effectively shared over two cables that may have significantly different resistance, so one will take more current than the other.
This type of certification is also available for a lot of C/C++ libraries, and it does help with the safety-case.
A pity a lot of projects then go on assume that the code using the library is, "by magic', also certified to the same level - the certification simply means it was developed to a started that is compatible with the safety requirements of a SIL 2 project.
Any decent critical system will assume that they can happen, but will introduce mitigations to ensure they are detected.
The "simplest" way is to use hardware so that the software can be written so that it does not have to worry about them. Whilst it can be done in software, it's not so easy as even (e.g.) CRCs only give transient protection (the data was valid when its CRC was checked, but what about when it is then used?).
Maybe, but it is non-trivial; simply storing multiple copies is not enough. For example:
bool f( int x )
{
static int c1;
static int c2;
c1 = x;
c2 = x;
return c1 == c2;
}
Many C compliers will optimise this code to always return 'true', as there is no way that the values of 'c1' and 'c2' can differ within the abstract machine that the language uses to execute the code; memory corruption is not considered by the machine.
If you could get some electronics into the beam at CERN, it's not going to survive.
Don't quote me on these figures, but the energy in the beam (at full power) is equivalent to something the mass of a large aircraft carrier travelling at 40 knots. There are dump tanks round the ring that are filled with water, and you do not want to be near one when it's used.
The sort of even that could have caused the Airbus issue is likely to be a (single) energetic particle causing a single bit upset.
The nominal supply voltage in the UK is 230 V, with a permissible variation of -6% to +10% (~216 V to 253 V).
The unsymmetrical tolerance was introduced to allow the UK and EU to "harmonise" and adopt the same supply voltage of 230 V when they were using 220 V and 240 V. Of course, no country has actually changed the voltage it uses.
This can have an impact on devices that were optimised to work at 220 V - for example, incandescent lamps that were designed for 220 V use in Europe will have a much shorter service life if they are used on a 253 V supply in the UK.
The voltage to my house was regularly going over the statutory maximum of 253v, causing my UPS to go into "trim" mode.
After about 18 months of pestering the supply company, they came round to install a data logger to see if I was right.
Guy told me he would need to switch off the power to connect his kit and asked if I needed to shut anything down first. I did, but as I went back in to do so, the power went off.
"Not me", he said, showing the wire that had just fallen out of the meter when a brushed it with his hand.
Turns out the metering company didn't do a very good job when they replaced the old meter with a "smart" one.
Again, that could have caused a fire. Luckily, there were no long duration, high current appliance in that house.
It was running at 500 amps, so a resistance of 10 mR creates 2.5 kW.
The specific heat capacity of steel is ~500 J/kg·K, and it takes another 275 kJ/kg to melt at ~1400 C.
Assuming the bolt was 50 g, then it would take ~14 s to get to the melting point, plus about 6 s to then melt (all of it, but it would fail before that happens).
That's one of the reasons I use a thermal imaging camera - makes it really easy to spot where there are bad connections.
I used to work on high power motors, and once spotted a wire swing out of a test rig and leave a trail of molten metal behind it - a 10mm steel bolt had been carrying about 500 amps between two cables when it was supposed to have been keeping them in direct contact. Those rigs used to loose a few kW of power in the connections even when they were tight...
I once had a manager who came up to me at my desk and, in an angry voice, said "You and I have a serious communication problem!".
This was news to me, so I replied with "You must be right, as I have no idea what you're on about".
He then stormed off and I stood up and started walking into the engineering lab - at which point I felt a hand on my shoulder and was physically dragged back into the office "to continue the conversation".
That incident was later cited as "a verbal warming" when he decided to formalise another similar incident.
My friend, who was a part time police officer, said they would be more than happy to support my assertion that this was assault!
Edited to fix a typo.