Why are university admissions so much more competitive now than they were just a decade ago?
"In 2010, USC's acceptance rate was 26.3%.
By 2015, the acceptance rate had dropped to 18%.
In 2020, USC's acceptance rate was 16%.
For the Class of 2027, USC's acceptance rate further decreased to 9.75%."
The correlary to that is that the "yield" goes down for all but the very most competitive colleges, so colleges have to build longer waitlists to ensure a full class. And more uncertainty for everybody.
You stopped short: why did they get more competitive? Because demand is outstripping the supply of desirable schools. Second tier schools need to improve their offering, and burnish their prestige, which takes time.
I don't think there's a satisfactory monolithic reason for this.
Maybe it's questions all the way down until, why do the common people need to get good jobs/hobnob with the upper classes? Because central banking policy of inflation?
The application process to many universities has been dramatically streamlined. It is in a students best interest to cast the widest net of applications for suitable schools— so naturally rates go down as the volume of applications per student is going up. It has also long been an open secret that universities are encouraging as many applicants as they can in order to reduce their acceptance rates to boost their prestige and rankings.
We need the actual data on university acceptance criteria itself (i.e. mandatory SAT score minimums) to truly assess the competitiveness of these schools over time. Raw % acceptance has become a diluted metric
“ Recent trends in college admissions include increased numbers of applications, increased interest by students in foreign countries in applying to American universities,[10] more students applying by an early method,[8] applications submitted by Internet-based methods ”
I can comment on Caltrain only, the location is available via the 511.org API.
I built a project that predicts how late trains are based on their current location vs historical averages and posts the data to Mastodon. https://caltrain.live.
Very nice, thanks for the write up :). I was thinking about how you could get route maps, without actually having to source shape files. Given you are tracking trains every minute, you could probably build up your own route maps by dropping a point every time you get one, and connecting them with straight lines. Over time the route would become closer and closer to reality.
G1000s are great, but it's disappointing that just about everything else has remained basically unchanged on GA aircraft for the last 30+ years. I hope that we will see some actual innovation in GA engines, airframes, and fuels soon. Fuels might be the most promising for the near-term.
The innovation/future exists, but is not evenly distributed. For example Diamond planes [1] have:
- The same G1000 (NXi)
- A composite body
- Jet-fuel burning engines adapted from recent Mercedes diesels
- Computer control of that engine through one knob instead of managing the throttle/mixture/prop separately like cavemen
- Crash testing, like impact absorbing seat structure and separately enclosed fuel tank modules that are unlikely to rupture instead of just filling the wing.
And a new one costs about the same as a new Cessna 172 that's been essentially unchanged since the 60's except for the G1000 like you said.
But "the same" is a pretty nice house in most of the country (~$600k) so everything is hand-built, so costs are high, demand is kept low. Commercial students trying to get their 1500hrs mostly just need the lowest cost, not the nicest or safest.
Another big area of innovation and also lower costs is in experimentals and/or light-sport. LSAs should be getting a lot more capable soon with MOSAIC [2]
Yes a new DA40NG is around $600k, as is a new Cessna 172. A new DA62 is well into $1.5m with the options you'd want if spending that kind of money, and a 2-year wait last I heard.
Meanwhile you could build something like a Velocity V-Twin well-equipped for under half a million. Paying for certified lawyering is expensive.
Nearly all the innovation in general aviation, particularly regarding situational awareness and safety, is happening in the Experimental category. Moving maps, synthetic vision/terrain alerts, ADS-B in and out, engine monitoring, airspeed-aware electronic trim control, fuel injection, electronic ignition, FADEC systems... yes, all are available on newer certificated planes for $,$$$,$$$, but much more affordable and accessible in the Experimental world. And, the builder/owner can install and configure everything himself. Garmin's latest G3X update includes the ability to use rudimentary set/clear logic signals to do if-this-and-that type CAS alerts, and configure your gauges to behave differently during different phases of flight. All owner-configurable.
I am interested to know whether or not the screws into the carbon hull, for example the monitor mount, might've had anything to do with the likely implosion. I have trouble imagining it being something so small, but I don't believe a whole lot of engineering rigor went into the design and build of this sub.
Coding Rooms | Backend & Front-end Engineers | REMOTE or Palo Alto/San Francisco
At Coding Rooms, we're building the first virtual classroom for teaching programming. Our customers are teachers, professors, and tutors from around the world who use Coding Rooms to deliver and manage their programming lessons online and in-person. Our problem domain means that we work on online IDEs, Docker infrastructure for running student code, and solutions for automatically assessing students' code.
This is (very unfortunately) true, and I honestly don't know people get it to work in larger school implementations. The idea behind BigBlueButton is spot on and ultimately I hope another FOSS tool can help to fill this void
This is a really brilliant project, and I love that it's open source! I'm certain that for tutoring, this is a much-needed self-hosting option! When it comes to schools, I'd be curious to know how it works in practice. Having supported a lot of schools in the past year using Coding Rooms with integrated Daily.co for remote learning, I have to say that the biggest challenges have always been related to just getting video to "work" with students. From the instructor's perspective, it's just so challenging to get 30 K-12 students on the same page with audio, video, input devices, and UIs -- not to mention stuff like just getting everyone signed in or using the right link. A system (perhaps hardware-included/based) that could do that would be ground-breaking for remote learning in my opinion. This is also the reason, in addition to school policies in some cases, that instructors feel like they have to stick with what they know in situations like this.
It's also worth noting that we started by integrating Jitsi, then BigBlueButton, and finally settled with Daily.co because stability, UIs, and support were such huge issues for us with the other solutions. While Daily.co doesn't do everything perfectly, I have to say their coverage of edge cases and support has been a lifesaver scaling our product for schools.
Yeah there is, the video chat is not P2P, everyone uses the SFU model where all your media goes through daily's server, so they have to pay for the instances and the bandwidth
Thanks for the kind words. Our goal is to asymptotically approach doing everything perfectly, and we really value feedback. So please let us know where we can improve and what you'd like us to add. Relatedly, sometimes we have sample code for features that aren't nicely abstracted into APIs, or which we haven't documented yet/well. So don't hesitate to ping us.
I can certainly vouch for you and your team's commitment to that level of perfection and use of feedback even during what I can only imagine was a truly unprecedented year of usage. The majority of our (relatively few) issues were small UI issues that confused students and instructors, but with the new beta of the pre-built UI I believe those have been addressed! Some of those less well documented patches have also been really useful in tuning performance for certain use cases!
Coding Rooms | Backend & Front-end Engineers | REMOTE or Palo Alto/San Francisco
At Coding Rooms, we're building the first virtual classroom for teaching programming. Our customers are teachers, professors, and tutors from around the world who use Coding Rooms to deliver and manage their programming lessons online and in-person. Our problem domain means that we work on online IDEs, Docker infrastructure for running student code, and solutions for automatically assessing students' code.
"In 2010, USC's acceptance rate was 26.3%. By 2015, the acceptance rate had dropped to 18%. In 2020, USC's acceptance rate was 16%. For the Class of 2027, USC's acceptance rate further decreased to 9.75%."
https://www.crimsoneducation.org/us/blog/usc-admission-rate/....