Back in the day, we used to run "tarpit" SMTP servers which looked like an open mail relay but ACK'd incoming packets only just barely fast enough to keep the remote client from timing out and giving up. The theory was that tying up spammer resources was a net good for the internet, as a sender busy trying to stuff messages through a tarpit was tied up waiting on your acknowledgement, reducing their impact on others.

Similarly, perhaps the right answer here is to limit the number of concurrent connections from any one network range, and use tarpit tactics to rate-limit the speed at which your server generate contents to feed the bot -- just keep ramping down until they drop off, then remember the last good rate to use for subsequent requests.

It would perhaps be interesting to randomly generate content and hyperlinks to ever deeper random URLs -- are these new crawlers more interested in some URLs or extensions than others? If you pull fresh keywords from the full URL the crawler requested, will it delve ever deeper into that "topic"? If their Accept-Encoding header supports gzip or deflate, what happens when you feed them a zip-bomb?

As a Fiver (not Fiverr), gotta love the folk bragging on having (or losing) 6-digit accounts. Way to make me feel old.

Given their "self-service" business model, why did Fiverr need 762 employees to begin with?

Theodore Kaczynski was caught because of poor Operational Security (OPSEC). He let his ego get the better of him, delivering a 35,000 word manifesto and insisting that it be made public.

He was caught only because he thought he was smarter than everybody else, leaving clues with each bomb and in his manifesto. Ultimately the Washington Post's publication of his writings caught the eye of researchers, and more importantly, his younger brother David, who turned him in for the $1M reward.

Definitely! Similar vibes to (but pre-dating) Trader Joe's "Fearless Flyer", right down to the cheap paper and retro graphics.

Visiting their original storefront in Chicago was one of my favorite excursions when I was young and in need of science fair inspiration or just "stuff" for one of my personal projects.

Pretty much all the B&M and online surplus electronics stores I used to buy from have faded away or moved to a purely eBay existence.

Generally those traditional crawlers are well-behaved, and will follow the instructions given in robots.txt, though not all follow suggestions like crawl-delay. And if not, they tend to originate from fixed source IP addresses which can be blocked or throttled by the site operator or their CDN.

Back in 2020 IETF released a draft document "RateLimit Header Fields for HTTP" providing rate-limit headers which well-behaved clients should respect.

They seem to be suggesting this is because students from "higher income" families were better able to adjust their behavior to work with the system, including lower absenteeism, turning in assignments on-time, and completing extra credit assignments. Oddly I was one of the lowest income students in my high school, and yet I managed to meet these minimal standards!

Oddly enough, these same behaviors (showing up, on-time-delivery, and going above-and-beyond) are also desirable skills for nearly any employer, thus "Grading for Equity" does a disservice to students by explicitly removing them from the grading equation

Looking at Joe Feldman's defense of GfE, the one thing he seems to get right is his suggestion that a student's grade in a given course should be most influenced by whether she mastered the subject matter at the end than how muchs he struggled with it at the beginning, middle of the term

The median is slightly more meaningful, as you are assured there is at least one sample in your set having that exact value,

They spent around $24.5M on tickets (assuming they were able to keep the 5% sales commission), took home $57.8M (the lump-sum payout of the $95M pool).

If one other person had won each winner would take home $28.9M, still a small net profit.

In college we explored replication the GPS algorithm, your final calculation would always give you two position fixes -- one in space, the other within the earth's atmosphere. Your algorithm would discard the orbital result and return the second answer.

Scratch that, reverse it, and you've got your answer for a lunar fix /s

The reason your self-constructed strawman suggests is just your personal biases showing, not the NRA's. Rather Castile's case and Biden's had one thing in common -- the person at the center of the case was a drug user. Also the NRA rarely has much to say about law enforcement shootings, regardless of the ethnicity of the victim

OTOH, " the 2A crowd" (which is a lot larger than the NRA) did actually have quite a bit to say about Castile and about Hunter, yet were mostly ignored by the major media.

Just get rid of the penny!

I wouldn't call that either compiling or interpreting. That's front-end lexical analysis and parsing, which for BASIC was not exactly a conventional parse tree, but close. It remained in memory, and that's what was read when you typed RUN. I'd have to fire up my old Apple ][ to be able to tell you for certain what the on-disk format of a saved BASIC program looked like, but it was probably stored in some hybrid of the tokenized format plus the REM comments. (I probably knew the answer to that 45 years ago.)

It wouldn't surprise me if they had considered compiling that tokenized data down to the same UCSD P-code that the Pascal system used, but AFAIK, neither of Apple's BASICs were ever ported to the p-System, although there was a different BASIC from UCSD that was available.

When I worked in the HP-UX file systems group, we had Hans Reiser come in for a day and talk about possibly including ReiserFS with HP/UX, either instead of, or in addition to, VxFS from Veritas. IIRC, this was about a year before he got in trouble and we realized just who we had been dealing with.

Our impression at the time was that he was no more eccentric than any of the other super-geniuses who worked with us at HP or who had built better mousetraps on the outside. We were hoping to swing a deal like we had with Veritas, where we got the VxFS code in source form, built it into HP-UX, and distributed it for free with the OS instead of as an extra cost add-on like it was with Solaris. In return, we received late-beta quality code, tested the bejesus out of it, fixed most of the bugs we found ourselves, then sent the fixes back to Veritas, where they could do whatever they wanted with them, which was to integrate them back into their code base nearly 100% of the time. Part of why this worked was because Veritas and our organization were physically located relatively close to each other, so face-to-face communication was easy and frequent.

We decided not to proceed with Reiser, not because he was a nutjob (which we had no way of knowing about), but because he wanted too much money, did not want to share the source, and because his core developers were in Russia and other countries and were contractors instead of employees, so they could disappear at any time. For someone who was going to be a critical third-party contributor, it looked too brittle for comfort.

Not impossible, especially in an OS environment (either monolithic kernel or microkernel and separate FS code) with well-thought-out VFS and I/O layers, but yes, very difficult. Even writing one from scratch and putting huge resources towards every known type of testing may still leave in some bug that doesn't manifest in the field for years or decades. It's especially frustrating when you're trying to hit the sweet spot of the fastest-but-most-expensive versus slow-but-cheap-and-reliable, like bcachefs does.