> Fair use is the same doctrine that allows a school to play a movie for educational purposes without acquiring a license for the public performance of that movie.
This is a pretty bad example, since fair use has been ruled to NOT allow this.
There have been several recreations of Emacs in Common Lisp over the years, including Hemlock, CEDAR, and Climacs. They never catch on, because they’re not compatible enough with the existing base of emacs lisp software. This also applies to Scheme, although it did get closer to compatibility once or twice.
This is also closely tied to why emacs hasn’t been made much more parallelized - fundamentally, the design of emacs is too close to “a giant ball of global state”, mirroring an early Lisp Machine. It’s a high hill to design around, and it will violate basic assumptions made by most medium-plus interesting elisp code.
Do not let this dissuade you from trying! It would be a great if difficult accomplishment!
It turns out that most fast food burgers are made from heavily processed meat that includes various preservatives and whatnot that make it noticeably less healthy than plain hamburger, unfortunately.
But it’s $24/month, isn’t it? Double that if they invited the kids, in-laws, or neighbors but they don’t use it regularly because those people don’t use Slack for business — which is the whole point of the loss-leader approach: get people used to using slack, get them in the app, and then get them to bring it into larger environments where corporate rates make sense.
This new policy degrades the value of slack significantly for a bunch of long-term, low-volume users. It will cause a few people to move to a paid plan, maybe, and it will certainly cause many to leave slack. The last part will make slack less sticky, less familiar, less generally used, and will save them some amount on storage, maybe? Have you priced out storage recently? I’d bet the cost difference for storage of short text snippets between 90-day rolling and 10k messages is:
- Smaller than the variance of MTBF of different batches of disks
- Smaller than the variance on overhead in block/page size for their underlying storage system
- Smaller than the variance of employee number for their 90th percentile large customers
In other words, they’re adding annoyance and losing users for peanuts.
This copy is hosted in (one of) the old homedir of Karl Ramm, who passed away not so long ago. I believe that he would be tickled and mildly but pleasantly surprised to find the link on HN.
To whomever posted it: thank you for reminding me of an old friend.
“TikTok is unique...” might be true, but then why is TikTok not the only company targeted by this royal edict?
When 17 year olds can “own Twitter” and Facebook just budgets 9 figures for data collection compliance fines, are you sure there’s a bright line distinction in the world that’s not “let’s all agree we’re not the terrible/hated/untrustworthy other”? Is the problem that another sovereign country could coerce a company to provide a service that Facebook provides for money?
FWIW, a common thing to do with a chromebook that has gone out of update service is to install either a non-Google ChromeOS like Cloudready, or just install Linux directly. Both approaches are usually easy and effective.
It doesn’t stop working at all. It stops getting free OS updates, presumably because there are many dozens of hardware configuration to manage, and having fallen into that trap once with Android devices, they didn’t want to immediately jump into the same pit a second time.
I don’t know what things are like in other parts of the world, but the US market has shown a preference for renting large-library access over buying individual works. There are exceptions, when the costs got skewed too far and the perception grows that the “large library” is actually a small library plus a lot of dross (see Cable TV), but even then it took decades of Worst-in-the-nation service issues, billing ‘hijinks’, and predatory pricing to convince Americans to start cutting the cord.
True. Now, compare the number of music albums produced per studio to the number of movies. We’re explicitly talking about paying per time rather than per work here; one movie versus one album is misleading and/or irrelevant.
There are a couple of these MDS vulnerabilities that also exist on AMD, but the vast majority are Intel-specific. Thanks not to say that people won’t find some on AMD, but a) they haven’t yet, b) there are at least some that are much less likely on AMD than Intel, and c) the greater number of cores and channels is likely to make practical problems even less likely.
Some of this also applies to ARM.
I don’t have any experience with NordVPN, but off the top of my head, it’s not hard to imagine someone looking at an internet backbone map, doing some math, and just paying for a couple hundred residential broadband packages. (Napkin speculation: maybe 30 major markets, round up to 50. Most markets have only a couple serious residential broadband providers (sadly, in the U.S. the average is below 2), so multiply that by 3 (a big over-estimate, but in some places it’s probably worthwhile to double coverage). Sprinkle the remaining 50 around the smaller markets to make sure you have decent access into all of the major backbone loops. Ballpark it at $50/each and you could blanket the U.S. major population centers for $1k/month.)
Technically possible, practically unlikely (you iOS phone would be a terrible router, giving the people who made the calculator app a slow, unreliable output node that would identify to the desired source as a lower-quality (mobile) client. In other words, it would be a bunch of extra work and risk for a worse result.
The theoretical model here is using people’s fast, stable (-ish) broadband connections to relay the connection. Even then, it’s only for the initial setup steps; once you get to the actual streaming data, nobody in this model wants to ferry those packets around.
(This part of) Akamaiks value prop rests on two things: they have servers all over the world, and they can figure out which of those are both functioning well and (network-wise) close to you. The former comes from infrastructure investment and management. The latter comes from collecting and processing data from all over the (network-wise) world, quickly. That last part involves building and updating a map of actual and potential internet traffic — not quite all of it, but for everywhere that your (prospective Akamai customer ‘you’) customers might be. Doing _that_ without owning ~all of the BGP routers in the world plus ~all of the local ISPs in the world involves a fair bit of probing, data gathering, and a bunch of math.
It’s likely that those headers don’t all get added all the time, for all the Akamai traffic, but instead are added selectively for key parts of the mapping process.
(Disclosure: I worked for Akamai Way Back When, but left the company many years ago.)
Nobody claims that NordVPN customer connections are being used for this; the claim is that they are probably using Oxylab, which is probably using Oxylab customer’s connections, which they are probably using via malware of some unspecified type, so the Oxylab users probably don’t know how their connection is being used.
While some of those leaps of logic are plausible, it’s a lot of guessing.
I would argue that the stronger reason is that Windows typically doesn’t enforce, allow, or contains any distinction between kernel code and user-space code. This happens because there is a gigantic gravity well of pre-existing code that predates and/or precludes real separation, and the costs of concretely losing that support was nearly always judged higher than the abstract benefit of adding separation.
One old example: I have an old friend who used to do sysadmin work in the Windows 2k days, and he would often get gigs by speeding up services by 5-15% within a day. His go-to trick was to log into the headless win2k server and disable the screensaver process, then wait.