That Java one didn't sound like a bug so much as a disagreement about what the proper response to a user-space application accessing unmapped memory should be.

It sounds like Java was using undefined platform-specific behavior (undefined as in, not specified in a document saying that is the intentional behavior) - unless it's common for user-space apps to treat invalid memory accesses as recoverable?

To me it sounds like a security issue - if an app can just start trying to access memory, it can get side-channel information if it can recover instead of getting killed.

Inconvenient? Sure. Annoying because release notes never contain that kind of detail? Absolutely. Bug? Depends on if the change was intentional or not.

That said - the deletion of information on iCloud absolutely sounds like a severe bug, and I don't understand why processes weren't followed to check before it was released. It's just sad - where's the engineering diligence?

I've worked on and delivered at least a dozen "waterfall" projects - and just like any other development model, there's no such thing as "pure" waterfall.

"Iterative waterfall" is a practical reality. I don't know why people aren't aware of this? Is it just not common knowledge? Is there just tribal stigma? Basically there's a lot of advantage to having just enough waterfall with just enough iterative/agile. Going all-in either way is generally problematic.

The counterargument is that if you're writing code before you're designing the solution, you're setting yourself up for trouble.

Why is it that every other engineering discipline designs their product first, then builds it, but software seems to have this idea of building it first?

Sure this is a reductionist observation, but my experience is that probably 80% of software problems are due to "write code first" instead of "think about it and design it first, then implement the design in code."

There's definitely a continuum of prototype vs design. Both a benefit and detriment of software is that it allows very cheap "prototyping" - the problem is the prototype is often then shipped as the product.

No other industry tends to ship prototypes to end customers, because prototypes are not robust.

I don't know why so many software folks - and software management in particular - are afraid of engineering discipline.

The dollar amount of that minimum wage doesn't matter. The question is - how many hours of labor at minimum wage (after taxes) does it take to cover rent? How many hours of labor at minimum wage (again, after taxes) does it take to cover food?

More difficult to factor is things like, probability of losing your home, ease of going on vacations, ease of changing homes or changing jobs, availability of transportation, flexibility of transportation (is it on a public transit schedule, or is it "personal" travel?), fraction of time spent on leisure versus getting necessities, education, local population diversity and acceptance of others versus insular/fear behavior, and those other things that are really difficult to cast in terms of wage because they often depend on personal preferences or have a large subjective element.

Apparently they were using some kind of expansion. I don't know why you'd need that much more (10x !?) than 105 terabytes to store 105 trillion digits, and that's if you're storing each base-10 digit using 8 bits, which is already an "expansion".

I think what Sweeny is missing is that the 30% (or whatever) commission makes for very low barriers to entry. Instead of having to come up with some large initial risk, you can release an app on any of these "high priced" storefronts with basically zero (a couple hundred bucks maybe?) for 30% of gross.

This is really reasonable for a small enterprise compared to traditional markets, which require a significant at-risk investment for initial sales.

I think what Sweeny's complaining about is that the 30% (or 15%) is actually painful for the big players, not the small ones.

I don't know why he thinks the commission for access to a market should drop to the marginal cost of bandwidth and transaction fees; this neglects the fact that there is added value in the network effect provided by the stores, so of course the fees would not drop to merely the "cost".

Also even using his example - 25% gross margin, which he says is infeasible, is still pretty damn good in most industries. If you can't run a business on 25% gross margin, you've got serious problems.

This is the one thing I don't like about the DMA: it is "limited" by market cap and the subjective definition of "gatekeeper."

Make the DMA apply to Epic itself; make it apply to XBox, Sony, and Nintendo; make it apply to Steam; make it apply to every auto manufacturer; make it apply to every PC out there (I should be able to put whatever BIOS I want on it), make it apply to any phone (why am I locked into the phone vendor's firmware?) make it apply to any manufacturer of electronics that "locks" your ability to change its software. TVs, appliances, the works.

So I haven't really used Windows in over a couple years... are there not already one-click taskbar or simple key-combos for all of those things that are in that list for what copilot can do?

Why would anyone even suggest that you'd need AI for those things? Aside of course from managers wanting to monetize everything?

Yes, texting and driving, just general "I can do whatever I want" attitudes, things like that.

I don't dispute at all that "merely average" self-driving would eliminate many crashes. I dispute though that self-driving tech would be affordable for enough people to make a big enough difference. To eliminate even 25% of road fatalities, you'd have to have sufficient market penetration to cover them; since those accidents are not evenly distributed across the entire population, you'd have to target specific geographic regions and/or demographics.

Which, again, are not technical problems.

Of course, if you address the social issues, like substance abuse and distraction, you get more benefits than just traffic-related. So why don't people spend money on that better rate of societal return? It's a question worth pondering, not dismissing too quickly.

For the record, I think self-driving tech is quite interesting, and I wouldn't mind seeing it deployed, because I'm starting to feel downright afraid to drive on US highways. The aggressive driving habits of people have definitely changed over the past several decades. I can definitely see self-driving cars reducing on-road stress.

Indeed - it's the fallacy of not understanding the difference between absolute numbers and proportions, among other things.

Yes, the absolute number of deaths per year due to vehicle incidents is a Scary Number. However, it amounts to roughly two deaths per day, in each state. This is not a "high rate" even though it's a scary total number.

Humans really are bad at understanding risk, full stop.

This doesn't even get into the fact that car crashes are not "random" (even though they are statistical) - so in order to get rid of them all, you have to address every special cause. Getting rid of impaired driving is indeed a good goal - and that doesn't even require "self-driving."

Consider that a few years ago, the annual vehicle death rate was only about 30k/year, and it spiked up by 10k/year recently. What caused that? It wasn't the cars, the cars are safer than ever before - what changed is people's behavior.

The problem is, it's really difficult to get convince people (especially in the USA) to change their behavior for the sake of others; instead they would rather spend cumulatively tens of millions per life saved to require back-up cameras in cars. And that was only to reduce something like 200 accidents per year. Can you imagine the total cost to society to eliminate 2000? 20000? The amount of extra computation power, specialized tech, and manufacturing capability, tradeoffs in reliability and repairability, increased barriers to entry in the industry, to get these marginal gains is massive.

We have already eliminated 90-95% of the causes of fatal vehicle accidents. Getting rid of that last 5-10% is challenging and expensive.

Dunno - if you've held both the chief engineer and chief mechanic positions and you don't ensure production processes are safe, I think you do indeed need to go.

At that level your sole job is to fight off the beancounters and make sure things are safe - in spite of the beancounters.

Probably played too much Horizon: Zero Dawn - the Frozen Wilds

A significant portion of the plot involves an AI tasked with managing the thermal characteristics of a volcano.

Indeed - the wolves didn't "develop immunity" to cancer. It's more likely that wolves with genetic immunity to cancer were most likely to propagate, so that of the wolves that survive, they, well, have survival-bias toward cancer resistance.

It's very unlikely that the radiation itself caused a mutation that made the wolves more resistant to radiation.

Don't forget we're now in the Brave New World where you can't even trust screenshots on the internet, or videos, or even phone calls, because there is a substantial probability they are generated, not recorded or an actual person on the other end of the comms link.

I was thinking the opposite: Since there are some folks on both Disney and Apple boards, I'd expect this $1.5B to come with a string like "stop badmouthing Apple".