upvote
I don't know the cause, but if you step back -- 13% of a single airline's departures are delayed. This isn't all that rare.

I suspect it's a good example of media bandwagoning and availability cascade.

reply
This the stance I'd usually take if it was just 2 or 3 incidents, but with 4 major systems being completely disrupted within 4 days is really odd.
reply
The United issue affected 13% of flights and was resolved in under 45 minutes.

This would not even be news if people weren't jumpy after last week's chaos.

reply
This. At $PREVIOUS_JOB people would call in the day of a release (a saas product) and insist something broke in a new way for them only for us to dig in and see it was an existing bug that people only noticed because they knew there was a release.
reply
You’re missing his point.
reply
It's not odd in the brave new world where any idiot with an LLM can pwn your company, and where any internal idiot with an LLM can push out a fix for that exploit that will break your company.
reply
Part of it is recency bias. For instance after a plane crash, the media tends to report on every single aviation incident following it that otherwise wouldn’t have made the news. We’re seeing the same thing play out here in the IT space.
reply
It seems patently obvious there are a series of cyber attacks going on. In the UK we have been told to prep food and water, which is obviously for a cyber attack, so the govt clearly knows more than they’re letting on.

They don’t want people to panic, they want to decide how and when to retaliate without it being all over the press.

It’s going to get weirder and it’s going to take a while to get through this.

reply
> In the UK we have been told to prep food and water, which is obviously for a cyber attack,

Agreed - and we've seen cyberattacks against e.g. supermarket fridges so it's clear supply chain infrastructure is at risk.

> so the govt clearly knows more than they’re letting on.

Do they? It's no great secret that many internet-of-things/SCADA/PLC systems have very poor security, and that a lot of it gets connected to the internet anyway. It's also well known that any organisation of thousands of employees will have someone fooled by a phishing e-mail from time to time.

We also live in an age with a lot of sabre-rattling, and a lot of war - Ukraine/Russia, Israel/Palestine, America/Iran.

reply
The cost and special abilities required to hack almost any system have plummeted because of AI.

As such, cyberattacks are no longer limited to people and organizations with unusual capabilities.

reply
My pet theory: LLMs make software development go a lot faster. Their code also has fewer defects per unit shipped than human-written code does, on average and when LLMs are used properly. But since the volume is up, the absolute defect count is up too, even though the defect rate per unit of code shipped is lower.
reply
Any thoughts on why the changes, now being made more quickly, aren't taking the systems to a more robust state? Is it just the same old "we're not paying off tech debt" complaint, now at hyperspeed, or is something about the speed creating new kinds of debt?
reply
Most software systems aren't being made more robust because robustness in most orgs is still prioritized the same as it has always been, they're just shipping product faster. You could ship product with fewer defects at a slower pace, but that's not Moving Fast and Breaking Things™. Market forces dictate the speed at which product needs to ship, and you have to balance that with the defect rate to be competitive.
reply
It’s almost code freeze time for many organisations, so people are pushing things out to close the year with something for performance review.

Also, if you flip a fair coin enough times, you’ll eventually get a streak that feels rigged or strange.

reply
deleted
reply
> Their code also has fewer defects per unit shipped than human-written code does

Nope, and here's why:

> when LLMs are used properly

Yeah, and when they (humans) keep failing to use something "properly" there is always "No True Scotsman" argument. Nothing new in this world, we had it before. FP, strong type systems, memory safety etc. Bugs keep coming though.

reply
Don't just blame the programmers, either. Management mandates like "all developers must use X number of token per week" absolutely guarantee that LLMs are not used properly by most developers.
reply
Airlines have major outages frequently, its a pretty fragile system
reply
Oddly enough they didn’t used to and used to have rock solid computer systems with paper backup procedures. So did banks
reply
Don't worry, it's gonna get worse.
reply
Im genuinely 50/50 on what the boring Occam's razor explanation is. Because obviously it should be "software be buggy sometimes" like duh. However, hypothetically: - let there country A and country C - A and C are caught in technological arms race involving technology that enables cyberattcks - A has a much more powerful version of the technology than C so C probably does not want a direct escalatory cyberwar - A large event occurs in which Cs apparent access to an important enabling technology for this arms race is revealed and the consequences likely involve A cutting them off - The leaders of A and C are also scheduled to meet at a summit within days of the large event.

So we are in a period between the large event and a summit. A period where C may feel compelled to make a point/insinuate consequences before a pivotal meeting. But also if they do anything that's obviously a cyber attack and traceable it could escalate and they might not win.

If suddenly there are a series of high profile embarrassing disruptive cyber occurrences with no clear shared cause during that interval impacting As government and key industry players in their tech banking and defense sectors, well, look. Occam's razor is still probably "software be buggy." But also like they're on their way to a summit literally this week which sounds like a really obvious time to amp up the pressure a bit no?

reply
The boring explanation is no one actually gives any fucks at all about reliability. They sometimes think they do and they sometimes say they do, but when the KPIs meet the roadmap, reliability reliably gets prioritized behind shipping more features faster. Fail fast. Agile. Ship the MVP and iterate. Etc. If you've read fight club or watched the movie and remember the bit about whether Ford will do a vehicle recall, they're essentially doing that but for making the system reliable. There's a very strong argument that they're actually right, because as painful and expensive as these outages are, they pale in comparison to the money that gets raked in from being the first-mover and establishing a moat.

That's not a fun thing to arrive at for people who want to build reliable and secure systems, but if that's what you actually want, get a job coding something that is actually EAL (or something similar) certified even to just level 1, because everything else is just lying to themselves and everyone else when it says it cares about reliability. No, an SLA isn't equivalent.

reply
Anthropic later today: apologiez everyone, that was our model. Just hyping our IPO in this grim post-Huang environment. Please forgive us and don't forget to invest.
reply
While we're on the subject, there was also the Jensen leak and immediate follow up of other victims.... all seemed quite odd on timing with everything else.
reply
deleted
reply
gutting teams with experienced people in favor of cheaper talent with AI will produce these results. It's nothing new. We saw this in the past with heavy outsourcing pushes too, the velocity of the change is just unlike we've ever seen before.

Combine that with offensive cybersecurity research now leveraging AI and you have a recipe for disaster.

reply