Lumafield sells CT scanners, so these posts serve double duty as advertising for their capabilities. Given how many times their previous posts have been shared I'm sure the ROI is great.
I think it depends what your goal with Duolingo is. I've had great luck using the app as smart flashcards, less so an end-to-end language learning solution.
1. We adopted chip cards quite late, so a lot of legacy hardware still exists - including payment terminals built into restaurant point of sale machines with a magnetic strip reader, instead of a separate card terminal.
2. Wireless card terminals are not common in the US, outside of things like Square or Toast readers. Readers are usually hardwired to the POS and aren't able to be "undocked" and carried around.
3. Outside of contactless, signatures are still usually required, and people are still used to having the receipt printer at the host stand alongside the terminal, instead of built into the card reader.
This is all slowly changing, and I honestly chock a lot of it up to stuff like smartphone payments (remember, we got Apple Pay before having contactless embedded in the majority of cards!) and smartphone-based POS systems like Square.
Dark Sky had a big focus on "hyperlocal" weather prediction prior to the Apple acquisition.
If you're a weather app, being able to tell your users "it will start raining in 35 minutes but pass in 55 minutes" is a huge value add over "80% chance of rain in the next hour". I found it to be quite accurate down to the 5-10 minute increment when I used to live in an area with frequent summer thunderstorms.
That's not true with most of the prominent examples you see. All autonomous vehicle fleets in California operate under the approval of the state DMV, and do things like publicize crash reports: https://www.dmv.ca.gov/portal/vehicle-industry-services/auto...
I would encourage your daughter to cast a wide net. Being technical does not always mean being an engineer or a scientist (computer or otherwise), and being an engineer/scientist doesn't always mean being technical.
Some of today's most interesting and exciting "technical" challenges have their roots in many disciplines. Cybersecurity can be tackled from a full range of perspectives, from the highly technical (e.g. cryptography) to traditionally non-technical disciplines like law, public policy, and design. The kind of deep thought our world needs on things like AI and ML needs people who are just as informed about the social sciences, psychology, philosophy, and economics as they are about computer science. Our ongoing debates about issues like content moderation or digital privacy need folks who understand how to think about people, including those who are at risk and vulnerable, and then translate that knowledge into the language of engineers.
And even though we've made progress on this front, our governments, courts, and legislatures are still running on a deficit of knowledge about tech, which is a whole different ballgame. (I'd encourage reading Bruce Schneier's site on public interest technology if you're interested: https://public-interest-tech.com/)
All this is to say: if your daughter wants to go the hardcore tech route and loves solving CS or software engineering challenges, more power to her! But I also hope that she doesn't feel limited or boxed in by the traditional definitions of the discipline.
Interesting question! However the OCR functionality in the camera (through Google Lens) runs a separate pipeline. Other bug reporters said Lens didn't have the issue when scanning QR codes so I doubt it would surface there.
So far, I've been able to reproduce a few transformations:
* www[0-9].example.tld -> [0-9].example.tld,
e.g. www1.nyc.gov -> 1.nyc.gov
* example[ac|co|gov|edu].tld -> example.[ac|co|gov|edu].tld
e.g. exampleac.uk -> example.ac.uk
My guess, as someone else further down the comments mentioned, is that some URL handling library is doing more than expected to its input. I filed an internal bug report referencing some of the public reports from a dev build of Android 12, so hopefully this will get triaged soon if someone hasn't already done a similar reproduction.
(I work at Google, but on nothing remotely related to Android.)
I've found the Whole Foods store-brand smooth peanut butter is a good happy medium between Jif and the organic stuff. It stays emulsified a lot longer than others, but has more roasty peanut flavor (and less sugar) than the cheap stuff.
It's a double edged sword especially in academia. People who are excellent at executing in their field of choice are not always good managers. This is a bit of a personal soapbox, but I think one of the most entrenched problems in graduate education is the assumption that someone who is promoted on their academic merit will also make for an effective advisor/lab manager/project director/etc.
Depending on what you do, you might have a natural gift or be a good learner for management skills—but you also may not. If you're not and shoehorned into it for tenure (or whatever other reason), both you and your students suffer.
A good football coach doesn't have to be great at playing the style that they coach their players for. Having worked with her before, Dava is a great leader and manager - which is I think what's needed for a position like this.
I remember a review for the Little Drummer Girl that said (paraphrasing): "This book has achieved the rare miracle of making you feel sympathetic to both the Israelis and the Palestinians."
> simple regression models outperform the fanciest machine learning ideas for this problem
This reminds me of a classic paper: "Improper linear models are those in which the weights of the predictor variables are obtained by some nonoptimal method; for example, they
may be obtained on the basis of intuition, derived
from simulating a clinical judge's predictions, or set to
be equal. This article presents evidence that even
such improper linear models are superior to clinical intuition when predicting a numerical criterion from
numerical predictors."
Dawes, R. M. (1979). The robust beauty of improper linear models in decision making. American psychologist, 34(7), 571.
100% this. There's a huge subtext of "I'd be comfortable with more stringent measures if I knew the welfare of my population would be assured, but sadly it's not."
This kind of reminds me of the retail co-op dynamic. Most people in the US are familiar with the Ace Hardware brand, where local stores get a branding and logistics boost from being part of the co-op, but stay entirely locally owned. That way, independent stores can function with the supply chain and purchasing power of a large multinational. IGA does a similar thing in the grocery space.
One interesting thing about the Swiss system is they've prioritized reliable connectivity and frequent service over high-speed service. They do have some higher speed rail, but nothing even as fast as their neighbors in France or Germany. I feel like we have a thing for high speed rail here in the US as an aspirational goal, but maybe we should be aiming for reliable regional connectivity as a first step. Austin to San Antonio, Dallas to OKC, an electrified/more frequent SF to Sacramento - all of that could make a huge dent in local travel patterns.
This feels kind of like the article equivalent of a political cartoon: sure, it's accurate and worth a chuckle if you squint at it from a distance, but it's been done to death and I'm unsure if it's a productive portrait outside of the instinct to say "lol sooooo true!"