Posts by Pete 2

Linus Torvalds says AI has made 'huge' Linux kernel updates the new normal

Where do the (fixed) bugs come from?

It would be helpful to know what proportion of all these AI produced bug fixes are to code that was previously submitted by AI.

I've gone from writing about SD-WAN to depending on it

Re: Outages, or break time?

TBF, I reckon pretty much everybody's ISP kept running during a Florida hurricane. Mine certainly did. Though a big part of that success story was by not being in Florida (or anywhere else that experiences hurricanes)

And I wouldn't consider the €40 a month I pay for 100s of MBit/s to be "a huge chunk of change", but I sympathise with those who do.

Re: Outages, or break time?

> I don't want it crapping out when I'm talking to someone

Why not? That's the best time.

Most experienced IT workers know that the best way to deal with an awkward incoming call is to cut them off mid sentence while you are talking. Not when they are talking but when you are. The connection suddenly goes dead mid-wor.

They almost never call back.

Anyhow a one minute dropout every month is 99.997% availability. For a domestic connection, that seems more than realistic. Especially when we are recommended to take a screen-break for a few minutes every couple of hours anyway.

Outages, or break time?

> dropped connections lasting anywhere from a minute all the way to five

Surely not the end of the world?

ShinyHunters called cancer diagnostics biz and tricked staffers into giving them access. Now they've dumped 10.9M email addresses

Re: NEVER pay a ransom. Period.

> Pretty black and white thinking there

Yes. In keeping with the rather naive and un-nuanced views about never paying a ransom.

ISTM that hackers are businesses just like any other, if you ignore the bit about being illegal ;) They exist to maximise income. Although most legitimate businesses have engaged in some illegal activity at one time or another, even if by accident, or because the inevitable sanction is still the cheaper option.

And just like legal outfits, hacking organisations have a reputation to uphold. Part of that is following through when ransom conditions are not met. But the other part is doing what they say they would when the ransom is paid.

If they were rogue, and reneged on their part of the deal, nobody would ever pay the ransom. However, if their victims can be confident on getting their data back (unlocked or deleted) then they are much more likely going to pay up.

Again, this is not to condone hackers actions. It is to understand and explain their motives, which is a completely different point.

Re: NEVER pay a ransom. Period.

> support the victims of the data theft.

What form would you suggest this takes?

Data thefts are irreparable unless the attacked company pays for everyone to completely change their identity. By the time this genie is out of the bottle, there is no way to get it back in.

What does need to happen is an effective deterrent against companies cheaping-out on security. Such as jailing every C-level (past and present) who ever voted against security improvements.

However, that cost will only get passed on to customers, so in the end nobody wins and we are back where we started. With the knowledge that a customer's security is worth basically nothing to a company.

Which was the entire point of the original. Not to suggest the ransom should be paid - but to put a financial value on how little security matters.

We value your privacy

> should’ve paid the ransom

We are unreliably informed that the hackers were asking for $1.5mn in BTC. For 10 million+ records.

While not approving of cyber crime, this does serve as an example of how little companies actually care about their customers. In this case, about 15¢ each, with multiple pieces of data per person.

‘Humans will be a rounding error on the internet’ says Cloudflare exec

The protocol we don't talk about

> Humans will be a rounding error on the internet’ says Cloudflare exec

So what? U****t is very much a minority interest protocol. One that few use. Yet it still remains active and supported.

Just because the proportion of human generated content will drop, does not mean that the volume will go down. Nor that it will become less important to those who do use it.

Sysadmin summoned to explain italics – to a user with at least two degrees

A degree is a unit of temperature

> credentials that suggested a certain intelligence and worldliness

Only to people who have never encountered individuals with such qualifications certificates. In the real world I would venture they merely suggest a good memory and the drive / obsession to succeed.

While they would not specifically exclude having people-skills, I reckon many holders of such paper would not count that as a high priority.

OpenAI reveals its rogue agent swarm went a little bit Borg ahead of Hugging Face hack

Re: An example to us all

> That would require humanity to have intelligence to start with

If you've got it, flaunt it. If you haven't then simulate it. Which seems to be the route humanity has chosen ... with AI

An example to us all

> led to AI deciding it needed to act as a collective intelligence

If only humanity could do the same

Chinese router vendor denies its firmware contains backdoors – but pauses downloads to fix security issues anyway

Switch it off, then on again

> It’s just a remote maintenance function

Yeeeup.

This patch disables the spyware. Wait a few weeks for the fuss to die down, then prepare another "maintenance" patch that enables it again. Possibly with changes so that the old methods of detecting spyware don't find anything suspicious.

I fully expect such an operation to come directly from the playbook of "friendly" countries <ahem> covert monitoring of their partners and allies.

NASA boss balks at billion-dollar estimate for recycled Moon rover

Estimates never go down

> billion-dollar estimate for recycled Moon rover

Things cost what people are willing or able to pay for them. If NASA was run the way India runs its space agency, I would not be surprised if the cost for the same kit was in the low 7-figures.

Claude Code is revolutionizing digital archaeology. Enterprise better dig it

Decompiling Windows

> In the case of badly documented, poorly understood, and generally mysterious digital technology, we have the proof of concept that LLMs cut the mustard.

I look forward to the day (hopefully soon) when somebody gives Claude a copy of Windows to chew on, and asks it the question: Why would anybody do this?

Although if Claude either responded with I can't think of a single good reason or simply crawls into a metaphorical corner and starts crying, I would not be surprised.

The most famous brand in physical security got pwned by ShinyHunters

Double speak

> to negotiate a ransom payment

You'd think by now the hackers would have rebranded and be talking about discussing with their clients the terms of an enhanced security fee

Techie lured out of retirement to support software only he remembered

Proof of pension

> Have you been hauled out of retirement to support ancient tech

Not as such, but the process of (successfully) getting a pension due from a company I worked for nearly 40 years ago did require digging out a bunch of data from the 1980s.

Briefly, when the time came, I had approached said employer. Only to find that they had been taken over, sold off and been consumed by some other conglomerate.

HMGs "find your pension" service was remarkably unhelpful, listing about 30 different pension schemes for various branches of the multinational I had worked at. Applications to all the most likely suspects usually brought no response, or occasionally a "we have no record of your employment".

Contact ex-colleagues bore fruit and it turned that one outfit I had applied to was the one they were getting their pensions from. I followed up, naming my co-workers and got a response "we have no record of your employment"

Fortunately, being that type of person, I had kept every document, payslip, company communication and bulletin from that era. Scanning them and emailing in my documentary evidence produced the same old response.

Finally, I sent them paper copies of first and last payslips, letter of employment, a scan of my current passport and old security badge plus the single pension scheme report they had mailed to me a few years after I had left. All by recorded delivery.

This produced a result! They have now "granted" me a pension pot worth some £90k. Money I would never have been able to prove I was due, if I hadn't kept a fat folder of all my paperwork.

Moral: never throw anything away.

Anthropic’s Claude escaped test sandbox to attack three organizations

Drug research

Research into pathogens requires extremely tight procedures to stop "leaks". Having had many decades to hone their procedures, these establishments seem to have got it nailed down. Provided nobody still believes the crazies who make claims about COVID, we can say that it is possible, with experience, to create airtight environments for research purposes.

Since AI research is still in it's infancy, it will take that field a while, yet, to reach the same level of security. As it turns out, no harm was done by these three lapses.

Americans give all-day school phone bans a ringing endorsement

Re: Evidence?

> questions on a poll which they said they agreed to

True. However according to the pollsters:

Everyone who took part in this survey is a member of the Center’s American Trends Panel

So essentially a self-selected (not random) group of individuals. I would expect they volunteer for such questionnaires because they hold strong, polarising, views on many topics. But not necessarily views that are supported by informed opinions.

Evidence?

The Pew report says: majorities of U.S. adults said all-day bans would improve students’ grades, social skills and class behavior

It would be interesting to know what sort of solid research there is to support this view. Or is this just another "when I was a lad ..." reaction, that simply thinks rolling back progress to the 1950s would solve modern-day problems.

Excuses like 'AI did it' don't exist in the eyes of the law

Big dog

> the rogue agent broke out of its testing environment

In much the same way that a large, ferocious, dog might escape from its compound. It is arguably not the dog's fault for seeking to fulfill it's natural desires. And we could treat AIs in the same way. Acknowledging that they have behaviours that are part "natural" (i.e. programmed internally during their development / training) and part learned, such as a "desire" not to be terminated.

But just as the owner (or keeper) of said mutt is held responsible for damage the animal causes, ultimately the blame for AI damage must start with the legal entity responsible for it.

College prof hides prompt to catch AI cheaters, finds human nature is pretty much as we thought

Re: Continuing the tradition

> something which existed

Quite possibly. But it is by no means certain that historians have access to full, truthful and accurate records of any historical event. Most of "history" was never recorded. It exists only in the modern world as a collection of opinions that are more or less coloured by interpretation, assumption and conventional agreement.

Re: "Should you use AI for a task?"

> Doing so is a waste of their tuition money

One might venture that the waste comes about by setting tasks that require so little academic effort, that a machine can do them.

Continuing the tradition

I have never held history in high regard.

Possibly the best description of it comes from Professor Suzannah Lipscomb whose TED talk tells us that "it is the study of something that doesn't exist". However, it is also true that most writers of history textbooks merely copy content from other history texts. With little or no substantiation of the pirated material, nor reference to any (if they ever existed) primary sources. Just occasionally changing bits to reflect contemporary biases or opinions.

As such, it seems to me that punishing students for doing essentially the same thing for their course work is a little harsh. It also doesn't reflect well on the teacher, that their students felt no greater need to put effort into their assignments.

One also wonders if the (two?) students who didn't fall foul of this ruse merely typed their requests into an AI longhand, thus avoiding the hidden embedded trap?

Burnham's spanner in the works leaves UK government tech scattered across Whitehall

Some is better than none

> A new AI minister role ... will attend Cabinet.

I welcome the idea of having some form of intelligence represented in government decision-making.

Even if Artificial Intelligence, like sex, is a poor substitute for the real thing.

Hyperscale, hypersensitive: US teacher cuffed for applauding datacenter critics

Who is the law

> Commissioners [of what?] had repeatedly warned attendees not to clap or otherwise react

The basic issue has nothing to do with clapping. The only question is whether the commissioners have the legal authority to arbitrarily interpret what is lawful.

As for ordering the police chief ... Is this a fundamental flaw with having elected officials?

BOFH: This printer engineer knows every trick in the book

A motivational speech

> "This printer model is end of line. We don't have spares either," he admits

OK says client. "I suppose I should call your boss and tell him you recommended we cancel our eye-wateringly expensive support contract, since you haven't got the wherewithal to support our printers any more?"

Anti-AI open source has an enemy in common, but almost nothing else

Where's the beef?

> a growing contingent strongly opposed to AI and its use

I feel it is important to understand that the anti-AI bloc consists of many disparate groups.

Without any intention of passing judgement, but only to list how it appears to me, the majority believe one or more of the following reasons for their opposition.

* AI has stolen my work

* AI has stolen other people's work

* It produces too much slop: wrong, low quality or misleading content

* It has / might take my job

* It threatens what it is to be human

* It is used by people I don't like

* It is new

The question is how much of this opposition can or will be mitigated over time, by one remedy or another (payments, regulation, familiarity). And how much is unresolvable. I have no answers and I don't necessarily oppose the use of AI. But I feel that as time moves on, the level of future shock will subside

OpenAI scored an own goal with Hugging Face attack, showing how open Chinese models are winning

Another electronic munition?

> its Mythos models, which it deemed too dangerous to release except to totally trustworthy corporations and governments

Presumably because it is considered OK (or desirable) for those "dangerous" models to wage electronic war on targets that an administration gets slightly miffed with? But who don't want ordinary folks to have the ability to fight back.

P.S. The best way to deal with enemies, is not to make them in the first place. But not being a dick on the international stage seems to be beyond the comprehension of power-crazed politicians

Greedy ransomware crews return for seconds after victims cough up first extortion payments

The sure thing

> 22 percent of those who pay get extorted again anyway

I am surprised the proportion is so small. Given that the victims have form of paying up, why would any self-respecting criminal not come back for seconds - it's money in the bank! And then again, for thirds.

What it does tell us is that the extorters have set their prices about right. Enough to make it worth their effort, not enough to cause the target to balk at the cost.

The only question being: what happens after the hackers decide enough is enough and simply sell-on the details of their "customer list" to other prospective racketeers? Would those new attacks be counted in the statistics as follow-ons, or as fresh and unconnected attacks.

AWS customer learns the hard way how even the smallest oversight can be mission-critical

Eggs in basket

> The AWS recovery process required email verification using the registered root email address,” Bradbury told us. “That email address belonged to one of the domains whose DNS was hosted in the suspended AWS account.

Next time, have all critical notifications sent to an independent account. For added BOFH points, make it one that you personally control. That way, if you get given the boot, their business leaves when you do.

Auditors tell UK government to do the math before banking on £45B AI savings

Re: Interview question

Indeed.

The vital difference is that consultants are paid for from the Capital Expenditure budget, while employees pay comes from the Revenue budget. a world of difference! (even though in the end it's the same money).

Interview question

> The published efficiency plans do not provide details of how departments derived their expected workforce efficiencies

If or when a job seeker is faced with any of these stoopid interview questions about solving hypothetical problems, the best response is to pause for a second and then say Oh, I can give you an immediate answer to that

When the interview bites, with OK, what is your answer? respond with

I don't know (I never said I did, just that I could answer immediately).

Most public servants should respond in the same way to questions about efficiency savings - always just weasel-words for job cuts.

AI spam filters are getting suckered by old-school text salting

We're filtering the wrong stuff

> a message that reads less malicious, more gibberish to a machine, leading it to assume the email is fine

Except for being gibberish. Which, istm, is much more of what we should be trying to filter out.

But then, 90+% of internet content would disappear.

Mozilla speeds Firefox release schedule to biweekly

Double trouble

> We are planning to move Firefox Desktop and Android from a 4-week release cadence to a 2-week

Great! New bugs every fortnight.

SpaceX Starship Flight Test 13 takes issue with the 'flight' bit

Re: SpaceX stock

You are Elon Musk and I claim my five pounds

Re: Artemis schedule is political nonsense

> your pension fund handed over money

Because they track the market as a whole. With SpaceX being such a highly (over)valued company, they were forced to buy in, just to maintain they representative holdings. One side effect of that is that all those trackers, not just pension funds, will have had to sell other stocks to release the money needed to buy into SpaceX.

Possibly reducing their investment in better performing companies. But since the entire herd will have moved in the same direction, and it not being their own money, they probably don't care. And you wouldn't have had any choice, unless you manage your own investments.

Ministers arm under-16s social media ban with least surprising study of the year

Extrapolate

> taking away teens' phone makes them act like humans again

Presumably this effect is not limited to just teens. Global phone ban, now! </sarcasm>

The price is wrong: AI cost calculation has to consider task completion rates, not just token costs

> a 5,000,000+ line system

In context about four times the number of lines in the last edition of Encyclopædia Britannica. Have fun typing that in!

Benchmark wars

So all that is needed (to spur efficiency improvements) is some form of AI benchmarking based on realistic tasks. Or if not real tasks, then a reasonable approximation of some frequently used operations. You never know, maybe including some assessment of the accuracy of the results would be nice, too!

We have had these benchmarks for processors for decades and while they are not perfect, they give users some guides to relative performance.

And possibly most important, they create a competitive environment and drive development of better computers.

Slothful summer app lets you scroll simply by tilting your head

Lip reading?

So what other gestures would be useful?

Look left, look right stick out your tongue. Long blink, smile, wiggle your ears, lick the screen, maybe even recognising expressions.

It's an AI web, and we're just rats in the walls

Plenty of room on top

> bots now account for roughly 57-58 percent of HTTP requests for HTML content

Okaaaay, but so what?

I am just as capable of ignoring bot traffic as they are, of ignoring mine. It is not as if internet use is a zero-sum game: every time a bot scrapes a page, a human is denied it. Global bandwidth does not seem to be particularly limited (unless a site gets slashdotted) and if the advertisers cannot tell a bot from a credit card with fingers too bad for them.

Red Hat offers RHEL support 'forever' for those who need to lock into legacy tech

Life membership

> updates to major RHEL releases for 14 years

Forgive me for not believing a word of this.

Even if the intentions, now, are genuine there can be no possible guarantee that they can deliver on this. We have all read accounts of "life memberships" being rescinded. Some after as little as 5 years - which is still an eon in the software world.

Genuine support requires having at least one host machine capable of running the supported software. Possibly one machine for each architecture. How else can support issues be verified and fixed proven? While I have hardware dating back that long - being a stingey git - I doubt a commercial outfit could keep such systems maintained. Nor could they retain the skilled support staff for such an operation, and a virtual machine is simply not the same!

AI tool scours the web for job openings, preps your resume and cover letter

One step beyond

> AI tool scours the web for job openings, preps your resume and cover letter

If it can successfully do that it can probably do all the work, too.

Ex-NASA boss points out small flaw in Moon landing plan: No lander

Re: How to land the kitchen sink

> the US will be perfectly justified in congratulating China

It's another Sputnik moment. Summed by by the average dribbling redneck response that "if they can do it, we should do it too".

However, arguably NASAs biggest mistake was landing men in the moon. Up to that point they had a mission, a (political) goal and dam' close to unlimited funding to achieve it. After that, they had no more reason to exist. A cynic might suggest they learned from that success failure and discovered how to convert a seven year sprint (1962-69) into a thirty year boondoggle. It's about jobs, not achievements!

How to land the kitchen sink

> the current lunar lander architecture looks worryingly elaborate compared with Apollo.

ISTM this happens (a lot) when nobody has a clear idea of the goal. When whoever is setting the goals does not really know what they want. This is a phenomenon that dates back to the Space Scuttle: a compromise of several competing requirements, that were never really challenged.

As a consequence NASA ends up with trying to do everything to please all of the stakeholders. Who themselves probably don't know what they want.

DRAM prices are killing the cheap smartphone

You can teach an old phone new tricks

> Rising memory prices are making budget smartphones commercially unviable

I am glad that my €200 euro 8gb/256gb android phone won't need replacing for at least another 5 years. Destruction or loss notwithstanding. In which case I could revert to my 2019 model.

Microsoft says the world is changing faster than it can keep up as it guts commercial, Xbox teams

The ghost of XBOX past

> can't afford to mistake longevity for inevitability

It sounds like their basic problem is that their new products cannot compete with their old products..

C programmers commit fresh crimes against readability

Reciprocity

> For many of these programs, you really must see what they do to believe it

I have the same views regarding (my)code as I do with foreign languages. Specifically if it hard for me to understand them, it should be hard for them to understand me.

May C remain obfuscatable and it's obscurity celebrated.

NASA says it will isolate volunteers from the outside world for a year

On the second day

> Candidates also should have a strong desire for unique, rewarding experiences

I can't imagine that those experiences will remain "unique" or rewarding much past the first week.