But there is far, far more in the world that just SaaS or internal only software. Sure, companies are generally shifting towards SaaS (in part because it allows them another way to combat piracy, but also to extract recurring revenue rather than one off purchases), but there's still plenty of software you can go out to the shop and buy. Most of that wouldn't exist without copyright laws, because those developers would have found some other way of making a living instead.

Let's say you write some software for your employer and they sell the product to several customers with a contract not to distribute it. However, the product appears on some pirate website anyway, and you can't identify who allowed the product to be leaked. Once it's there, your contracts are worthless as nobody who downloads it from the pirate site will be bound by your contract. You need copyright laws to prevent subsequent distribution.

Without copyright laws, you'd then end up in the situation where people who paid for the software then are bound by the contract and would have to keep paying, and those who just downloaded it from a pirate site would be completely fine legally and could use the software with impunity. That would create a disincentive for anybody agreeing to buy the software and entering into a contract in the first place, essentially killing the industry.

You might not see a problem with that if you believe in Open Source, but most licences require copyright laws to exist to protect the software freedoms so that people don't just take the software and close source it.

Looking at London, the top rated place is Soho. Sure, a lot of people do choose to live there, but when I moved to London, I'm so glad I had to walk across central London on a couple of occasions before I actually found a flat. I went through Soho once on a Friday night and realised there was absolutely no way I'd ever be able to sleep at night and ruled it out. The next highest place is Seven Dials, which is fine for a night out, but again I wouldn't actually want to live there.

I actually chose to live about a minute walk from Hyde Park and 5 minutes from Paddington. That scores 89, so not much below the 96-ish of those top 2, but it's around 30-45 minutes walk from both of those. I think it only scores so highly because of the food options on Edgware Road, but there's not much else to do in that area, certainly not within 10 minutes walk, and I would frequently walk into more central London to do things. My cut-off before using the tube was 60-90 minutes walk each way depending on my mood, and would only use the tube on about 10% of days and averaged probably 90 minutes total walking on normal days. I guess my idea of walking distance is much higher than the 10 minutes suggested by the article. I'd definitely choose to live there again though if I went back to London.

EDIT: I've just looked at where I live now (a long way outside London). My current place scores 38 with only 1 in nightlife (a pub) and zero in all the other categories. Ironically, it gets the same score of 38 as an adjacent tile which somehow gets 0 in all the categories even though it's closer to the pub. Another adjacent tile scores 18 with basically the same stats, seemingly losing because it has fewer marked parking spots and no park (even though that tile is on a massive area of common land, hence why it doesn't have many facilities). The centre of the town, all the tiles score 96, which is true there are more things to do, but there's very few nice places to live in any of them. The whole thing seems a little odd to me, and I certainly wouldn't use this to decide where to live.

Most applications don't really care about their logical address space, and also most applications probably don't actually need significantly large address space - so restricting everything to 32 bit is probably bad, but there should be ample address space in 64 bit for many applications to coexist largely in the same address space. Certain applications like databases that want to memory map files don't fit this model, but many of today's applications do.

Since the introduction of TLB tags, using segment registers as selectors does largely support this model and allowing protection between processes without requiring TLB flushes on each swap, so I don't think it's a bad compromise, except that 12-bits for processes isn't really close to being enough.

In the UK at least, people usually take them to a second hand / charity shop, who sort through them and send the valuable ones to auction (typically early editions, 100+ years old) and then either sell them themselves (for recent books that are easy to get rid of) or sell them to specialised second-hand bookshops.

Most of the specialised second-hand bookshops rarely throw books away, usually if nobody buys them after a couple of years they end up in the extreme discount piles (20p, 50p etc) and probably only trashed if they still don't sell from there.

But, anyway this was on SunOS and specifically some Sun4c machines with a PROM password. I discovered a couple of new machines were put into the labs without a PROM password, so you could press STOP-A and get into the forth debugger, and read/write system memory. This seemed interesting, but I wasn't sure what I should be changing.

Then one day, I was reading some documentation, and there was a system call [1] that returned a pointer to some information about the current process (or maybe it was for a specific pid), and one of those fields was marked as reserved. I discovered that the field seemed to contain some kind of pointer, but I couldn't de-reference it, so I guessed it was something in kernel space, probably the information about the task needed for the task switcher. This also seemed interesting!

Using the PROM monitor to dump memory starting at that address gave a lot of interesting stuff, and at something like offset 22 there was a pointer to another structure and dumping the memory at this one showed my uid and gid, IIRC the uid was offset 6. Sure enough, overwriting the uid changed the euid of my process.

So, I could reliably get root on any machine without a PROM password by running my program, using the PROM to read the contents of the address it printed out, add 22, use the PROM to read that address, add 6 and write a zero to that address. (Assuming they are the correct numbers after 30 years of brain rot).

So, anyway, I had a bit of fun locally, and noticed that there was a memory mapped file that let you read the PROM contents. I then decided that I wanted to examine this on a machine that had a PROM password set, but this was harder than I expected. My initial plan was to rsh into the other machine as root, but that was forbidden by policy (and this would have been logged). I then thought about creating a setuid script with my new found root access, but that was thwarted as each machine had NFS mounted with root-squashing and their own local root partition. So, I decided I needed to create an additional "system" user in /etc/passwd with a password I knew (because that owned the binaries and wouldn't have root squash). The specifics of what went wrong I forget now because things rapidly got stressful so a lot of it became a blur, but the long and the short of it was that I managed to completely delete the NIS password file on that machine and so I couldn't log into it any more to attempt to undo my damage. However, I discovered that the non-NIS passwd file was still intact, and so I had to use my previous PROM trick to hack the "bin" user on a different machine, which allowed me to rsh as "bin" into the machine I'd just trashed (and that would definitely have stood out in the logs). Once I had a shell as "bin" on that machine, I was able to use the PROM trick again to upgrade that to a root shell and undo the damage I'd done before.

Some time later, there was some other root exploit knocking around, and I used that on a machine with a PROM password set, and discovered I could easily read what the password was just by reading that memory mapped file. As luck would have it, every machine had the same PROM password, and so from that point I could always have gotten root if I wanted it, but was always too worried about causing other damage that I never actually used it.

However, when I later became a sysadmin at the university, the other sysadmin was very confused one time when I did the same hacking root trick after he'd managed to mess up something similar on the root partition of a couple of machines and he thought he'd have to reinstall them all!

I just did STOP-A and entered the PROM password that he'd never told me, and did the same thing I'd done before and just said "yeah, we should probably change this password as it's the same as it was 2 years ago". The process of changing the password was annoyingly fiddly as the PROM password change tool was designed to only work on the console, which is why they'd never changed, so I wrote something to pretend to be an actual TTY so I could update all 100 odd machines remotely via rsh instead of logging on at the console of each of them. Fun times!

[1] I have some recollection it was an ioctl, but I'm a bit hazy on that now as it was over 30 years ago. I suspect it was the SunOS 4.3 version of the prpsinfo structure described in this: https://www.typewritten.org/Manual/Sun/SunOS/5.1/SPARC/man4/...

TBH, if using quarter here is just shorthand for "a decade", so it might be easier just to say plainly "I spent over a decade in each of A, B and C".

I was told then that we were very lucky to have that projector as there were only around 4 or 5 cinemas in the country with 70mm projectors (other than IMAX), and I can only imagine it's gone down since then. That said, it's hard to find information on it now - the only list of 70mm cinemas I could find had 5, but included IMAX and not my university cinema even though they showed Interstellar in 70mm last year so probably are just not listed because it's not open to the general public.

Oh, thanks for pointing that out. It's alluded to in the article, but I hadn't twigged that they'd increased the size of the frame. Yeah, that definitely means there's a lot of scope for just massively increasing the height of the screen whilst keeping it wide, and so it's definitely going to be an issue for a lot of cinemas to fit it in without just shrinking the image size substantially so the height fits, at which point the extra resolution would only make it look less grainy.

The choice of unusual film standards has always been to make the film a bit "special" when viewed in a suitable cinema.

The original format 1.33:1 was simply using 35mm wide film (originally 1.35") and after putting sprocket holes on each side, there was 1" of usable width, and someone decided to make a frame 1" x 0.75". The projected image was just scaled linearly.

When they started adding audio, the image was shrunk slightly to 22mm x 16mm (the ratio of this image was modified slightly to 1.375:1), and the audio track was just placed between the right hand edge and sprockets. They could use the same projector technology, just with a mask between the film and the lens so that the audio track didn't appear on the screen, and an optical pickup head for the audio that was placed below the mask and lens.

As an aside - a silent movie projector has a complicated gear system that advances the film 4 sprockets in a jerk, once every 24 times per second, while before and after that it is fed smoothly (so you have a "loop" of film above and below to take up the slack). You also have a shutter that closes for half the frame to hide the film when it's actually moving, and open when it's stationary. When the audio passes the audio sensor after the bottom loop, it's moving linearly, so the amount of light that can pass through a narrow slit is proportional to the voltage sent to the speaker. Later when stereo was added to films, they just fit 2 audio tracks into the space of one, so mono films would work on stereo projectors and vice versa. And even later, Dolby surround was encoded as a dot matrix pattern in between the sprockets (actually, it was just a film ID and timecode that was encoded) and the audio itself came from the 3-5 CDs that came with each film. But I'm getting ahead of myself.

These two standards stuck around for quite a while and then some of the later experimental TV systems started also using this ratio, probably to making filming non-live TV possible / easier as they could use the film process and show it on a TV using a telecine without wasting any of the image space.

Once TVs started to become popular, directors started wanting to make their films more special, something that couldn't be experienced the same way on a TV. And so began the arms race on picture formats. For a while 1.66:1 was popular, which used the same projectors as before but changed out the lens for an anamorphic lens which squished the image to make it wider. And soon, 1.85:1 was standardised (with a different anamorphic lens and mask). This was known as "flat" and was used by the majority of films. Some of the higher budget films were produced in 2.35:1, known as cinemascope or just "scope" and so now most cinemas had 3 sets of masks and lenses - square, flat and scope - and could show pretty much everything with a pair of projectors just by swapping the lenses. A few special aspect ratios were tried, but these were usually limited distribution to specialised cinemas, and cropped to a standard size for regular cinemas. In this era, usually the curtains could move to the correct width for each film, or fully closed.

70mm film has been around for quite a long time, but never really caught on as it was expensive to shoot - non standard cameras, roughly 4 times the area of film (and so at least 4 times the cost, but more because demand was lower), and limited where it could be seen as most projectors were 35mm only. It was possible to get projectors that had a swappable gate so you could also show 70mm film, but they were relatively rare - probably less than 10% of cinemas had these. The aspect ratio was almost always 2.2:1, and so while it was narrower than scope format, people would pay extra for the 70mm experience because of the higher quality (and less grainy look).

In the 1990s, TVs started shifting to 16:9 (1.77:1) which was a half-way compromise between improving the quality for films (1.85:1 scope doesn't need to be cropped much or only has small bars top and bottom) and leaving masses of unused space for older 4:3 TV broadcasts. However 16:9 was never used for cinema, and more directors switched to 2.35:1 for films after that to keep the "better in cinema" quality.

So personally, I don't think 1.43:1 is really any better for 70mm (after all the film is so much bigger, the quality is already improved), rather it's almost certainly been chosen so that it looks different, and much better at cinema. Just as films shown in "letterbox format" on TV with half the screen black looked bad on a 4:3 TV, now pretty much every TV is 16:9, watching a 1.43:1 film on it will have big black bars on the side and detract from the experience. The cinema will be the best place to watch this.

So, IMHO, it's all just a marketing gimmick aimed at those that want to pay a premium to see a film as the director intended it.

I'd speculate that the regular payments on annuities would therefore be higher than tontines because insurance companies would be able to factor in average life expectancy to the pricing more easily, rather than having to budget for possibly all the money continuing to be paid out to just one remaining person who happened to beat the odds and live well beyond 100 instead of the average around mid-80s.

Translation: "we won't be able to monetise it"

"We should not sell powerful chips or chipmaking equipment to China"

Translation: "we can buy them cheaper when there's less purchasing competition".

"Distillation is a much more compute-efficient process than training models from scratch. It allows China to build much better models than its number of chips would ordinarily enable."

Translation: "we've been outsmarted and lost our advantage because their way is faster and cheaper"

I think it's a bit rich to complain about distillation, when they been plundering the internet for years for copyrighted works to train their models on without permission.

A summary of some of the points early in the video: in 2025, China added 300GW of solar and 100GW of wind power. In comparison, between 2015 and 2014 China added a total of 78GW of coal power capacity, and for 2025 onwards an additional 161GW has been proposed or reactivated. So, clearly just one year of additional renewables capacity far exceeds the additional coal capacity over multiple years.

Around 5:25, the video then talks about the difference between capacity (maximum possible power) and generation (the amount actually produced) and another chart at 5:52 shows that even though capacity has increased by 6% from 2024 to 2025, actual coal power generation decreased by 1.2% over the same timeframe. So even though the headline is that China is building more coal power stations, in practice they're to cover peak load and to provide energy security and the actual energy production (and hence) emissions is decreasing. FWIW, the same chart shows a similar global trend.

[1] https://www.youtube.com/watch?v=qde4r4_uMU0

In that case, whilst it might be a derivative work, the original would be unavailable to be copied so it would effectively reset the copyright clock. This is effectively a parallel to the case in UK copyright law where an unreleased recording is covered by copyright for 50 years, and if released at any point in those 50 years gets covered for 70 years from the date of release.

> Copyright should last ten years, at the very maximum 15 years, after publication. This is an entirely solved problem.

It is the complete opposite to an entirely solved problem, because not everybody agrees how long copyright should last, or even if it should exist at all. You might think your solution would make it a solved problem, many others would disagree. Governments similarly believe that the copyright term lengths they have enacted into law are correct, and that the lengths enacted by other countries are wrong, and maybe people (including yourself) disagree with them.

As a consumer, I'd prefer copyright terms to be shorter in general, but as someone who's spent around half their career working in games development, and on the kind of games that need a team of 100+ and so need heavy investment and protection in order to recoup that investment, I also support the idea of copyright in general. While my specific area probably doesn't need copyright terms longer than a decade, I can also see that in other areas these lengths are justified - I still listen to older bands, and see it as entirely fair that e.g. Pink Floyd should still receive money if I want to buy an album they made 60 years ago. Similarly, if I want to watch Wizard of Oz, it's absolutely fair that MGM should still get a chunk of royalties from a streaming platform. That money is going to fund the films of the future.

This is fine for transactional work, where it's very clear - you do/make something and sell it, perhaps a carpenter selling a table or whatever.

It completely fails if, for example, an author spends 4 years unpaid writing their latest book, and after the first copy is sold anybody is free to duplicate the words from that book without giving the author a penny for that work. One of the specific reasons for copyright is to ensure that the author is granted exclusivity over sale of the works, such that they can earn a fair wage to compensate for the 4 years of their life without any income at all. How us it unfair that if that author lives another 50 years and people are happy paying for the book for the next 50 years, the author should continue receiving money for it, rather than just some unrelated party whose only skill is being able to scan and reprint the book for basically the cost of the paper?

It's also fair to allow copyright after the author's death, because their children also should be allowed to benefit from their parent's work too. In most families, the parents want to pass on a legacy to their children. Why should an author be any different? Or are you saying if someone owned property that provided a monthly rental income, on their death the children shouldn't be allowed to inherit that property or the income stream?

If you say that the parent needs to earn all the income from the novel themselves and only that can be passed to the children, what if someone has spent 4 years writing a novel, and is killed in a traffic accident a day after launch day? Should their children not be allowed to profit from their parent's work just because they died before the income stream was realised? What about the 4 years of hardship the entire family endured?

Corporations just extend this thinking, because legally a corporation is considered to be a legal person. If you're making a film that requires lots of capital and many participants, there's a lot of investment in that production. In this case, it makes sense that the individual artists sign over their rights to the work they do for a salary to the company (if they don't like a paid job, they can always go and create their own works or set up their own company, etc) and equally, the investors need some time to recoup the investment they made. If we didn't permit this, then simply put, the vast majority of films would never have existed. Society would be far worse off.