https://github.com/grahame/church-calendar

That compliance aspect seems like one thing that pushes us towards centralised architectures for social media, but I'm guessing that AI models to screen images / videos are pretty widely available now and cheaply deployable?

Comments URL: https://news.ycombinator.com/item?id=42344537

Points: 3

# Comments: 0

"My dear Doctor, they're all true."

"Even the lies?"

"Especially the lies."

Comments URL: https://news.ycombinator.com/item?id=41223907

Points: 42

# Comments: 12

It whinges if it thinks I'm speeding, but it's often totally wrong on the speed limit: reading the wrong signs, holding a lower limit when I've turned onto a road with a higher limit. I find the alerts extremely distracting, and unfortunately they use the same sound as alerts I might want to know about. It also makes that same sound every time it reads a new sign. And then it feels the need to claim I'm speeding all over again, even if the limit hasn't changed.

I turn it off most times I get in the car, although there's no shortcut for it, so it's just a constant annoyance: the car won't remember the setting.

The boundaries between religion, culture, language, myth and storytelling, reflection.. these are all extremely porous and vary widely. It looks to me that the researchers in the article are aware of this.

Brent Nongbri's book "Before Religion" is excellent on all of this.

(Full disclosure, I'm an Anglican [Episcopalian in the US] priest)

Comments URL: https://news.ycombinator.com/item?id=39981034

Points: 945

# Comments: 104

In 2013 this became important: in Western Australia, we had a very close senate election. There was one critical exclusion in the count where just 14 votes determined the outcome between two candidates. Obviously the accuracy of the counting software was key; it actually crashed when they were doing the count. They restarted it, but that shook the confidence of a couple of people I knew, and so I decided to write my own software to verify the count. I knocked it up in two all-nighters: https://github.com/grahame/dividebatur

Of course, I/we were fortunate that the electoral commission was forward-thinking enough to have published the data required to fully reproduce the count.

Some open government folks later on used the existence of my software to try and get the electoral commission to release their software system under Freedom of Information laws, so that it could be verified. I was quite amused when the commission alleged there was no way I'd done it in two nights. I had; but of course, what I had was a Python implementation of the count, not a fully-fledged electoral management system like they had.

Later on in 2017-18 we had a constitutional crisis[0], as various senators were found to hold foreign citizenship and thus be ineligible to hold office. Those ineligible senators were replaced by running a count-back of the vote, with them excluded. I happened to be the only person who had a system that could work out the results ahead of the electoral commission, so I had quite an exciting few weeks providing the media with predictions on who would take over the various seats that were lost.

Now there are better and more robust systems that have followed mine, but I must say I was quite happy with this two-day hack!

[0] https://en.wikipedia.org/wiki/2017%E2%80%9318_Australian_par...

At a technical level both types of call are a 64kbps data connection, with just a flag to discriminate between the two. All the major suppliers (Cisco, Nortel, ...) had software options for their ISDN gateways to tell them to set the voice flag when making an outgoing data connection, and to treat voice calls as data calls when they were received.

This was extremely common and Telstra never seemed to do anything about it. Just a funny memory that this article brought to mind.

https://github.blog/2021-05-13-video-uploads-available-githu...