Article note: "A study found that novel, personalized, and interactive interventions improved student outcomes; why aren't one-way impersonal corporate-speak messages broadcast by a robot on a now-saturated advertising channel having the same effect?"
Article note: This is an experiment I've been wanting to do for _years_, I love that someone who isn't me spent the thousand-odd dollars on it. I would have liked to see an airbrush compressor in the test set, but small detail.
Those aquaculture pumps are apparently as promising as they look.
If there’s one thing that continues to impress us about the Hackaday community as the years roll by, it’s the willingness to share what we’ve learned with each other. Not every discovery will be news to everyone, and everything won’t be helpful or even interesting to everyone, but the mere act of sharing on the off chance that it’ll help someone else is really what sets the hardware hacking world apart.
Case in point: this in-depth analysis of laser cutter air-assist methods. Undertaken by [David Tucker], this project reads more like a lab writeup than a build log, because well, that’s pretty much what it is. For those not into laser cutters, an air assist is just a steady flow of air to blow smoke and cutting residue away from the beam path and optics of a laser cutter. It’s simple, but critical; without it, smoke can obscure and reflect the laser beam, foul lenses and mirrors, and severely degrade cut quality.
To see what air-assist methods work best, [David] looked at four different air pumps and compressors, along with a simple fan. Each of these methods was compared to a control of cuts made without air assist. The test was simple: a series of parallel lines cut into particle board with the beam focused on the surface at 80% power, with the cut speed slowly decreasing. It turned out that any air-assist was better than nothing, with the conspicuous exception of using just a fan, which made things worse. Helpfully, [David] included measurements of the noise levels of the compressors he tested, and found there’s no advantage to using an ear-splitting shop compressor over a quieter aquarium air pump. Plus, the aquarium pumps are cheap — always a bonus.
Not sure how to get up to speed with lasers? Laser Cutting 101 might be a great place to start.
Article note: Oh ed-tech carpetbaggers, they never miss an opportunity to rent-seek.
The maxim that the greater the separation between the people making purchasing decisions (district bureaucrats or deanlets) and using the software (instructors, students), the less suitable it will be.
Article note: Who would have imagined that buying dying tech companies then jerking around their remaining desirable brands would result in selling them a couple years later for half of what you paid...
Verizon has sold its AOL and Yahoo properties to Apollo Global Management in a deal said to be worth $5 billion, about half of the nearly $9 billion Verizon originally paid for the pair. Verizon will maintain a 10 percent stake in the company, now known as Yahoo and led by CEO Guru Gowrappan. The deal, which includes Verizon’s ad tech business, was heavily rumored over the last week and is still subject to closing conditions. Once complete, it’ll bring an end to Verizon’s troubled experiment with media production and advertising.
Article note: And again. Because once someone looked into the absurd nest of generations of half-baked hacks to make the numbers go up inside a modern high-performance commodity microprocessor, the shit-show is going to keep unraveling until the parts perform worse than if none of the speculation and hidden caches and such were there.
The assumed environment commodity computer hardware (...and software) was designed for was not multi-tenant (VM/Cloud), and was not 'automatically download and run random code from the network' (browser-as-runtime). It's possible to design computers for that, and at one point a bunch of larger systems vendors did (...and IBM is the only one still hanging on), but we're dozens of generations into lines that were designed as single-user detached systems then outgrew themselves.
Article note: I find the discussions around the merits of shared libraries really interesting, largely because there are a population of different answers all of which have clearly good and bad points, and the industry keeps iterating over them.
Some of the problem is clearly caviler dung beetle programming, but there is an actual problem in there.