@nemeciii agreed on the ACC, I haven't tried it in a city yet but I don't really trust it in heavy traffic on the highway. When the road isn't busy it's great though. Good luck with the apartment charging
Local gov't is funny sometimes. We needed our front steps replaced and the company doing it started on saturday but couldn't finish because some of the material they had weren't the right size. They came back today and right after they showed up the town inspector came and made them leave because they didn't have a permit for the work they were doing. Now I'm stuck with partially finished steps until the town approves the permit, which are definitely less safe than the broken steps that were there before.
@andrioid I'd be interested to see how this would work from a privacy perspective. Unless the classification is done client side, text data that the user hasn't posted is sent off their device. Here I see two cases:
1. Store the text even though the user hasn't explicitly made it public- this is obviously had from a privacy perspective.
2. Don't store the text. Better for privacy, but over time your training data becomes more positive with time, which presumably increases classification errors
@am1t there have also been several academic studies that show it's possible to combine several (sometimes publicly available) anonomyzed datasets that together are pretty good at de-annonomyzing individuals
@elmiko I recently got myself some recording equipment and have started recording music (nothing realeased yet) and I'm also working on getting a blog set up. I think being in more of a creator mindset will help
And yes, the irony in posting about this on social media is apparent to me ¯\_(ツ)_/¯
Today I was finally able to put into words something I've noticed about myself recently. I'm a constant consumer - podcasts on my commute, Reddit during downtime, hackernews while builds are running at work, whatever on my phone before bed, etc. I've found myself unable to simply sit without any stimulation. Today I drove home from work with no podcast, no music, with the windows down, clearing my head. I think that's a positive change. Now that I'm conscious of it, I can start to correct it.
since this is coming up again, i feel it's important to stress the following facts about privacy on mastodon:
- DMs are stored in plaintext in the database
- yes, this means admins can access them
- it also means they will be accessible if your instance DB is leaked
- however: this is the same as the situation on any other mainstream social media site
- at the end of the day you should make sure you trust your admins
- ultimately: don't use mastodon for privacy-critical messages
@natecull yikes. It's really easy to forget that these people are everywhere. I definitely agree that it seems like YouTube's recommendation algorithm is more sculpted (for lack of a better term) than they'd like to admit. I'm curious how much of it is actively pushing an agenda vs. just promoting controversial content to keep people on the platform
@colincornaby I'm sure it's possible (whether it's a good idea or not, well there's a reason we're not on Twitter, but that's beside the point) but I suspect there's something in the API user agreement that would prevent someone from doing that kind of bulk republishing. The license agreement references "copying a reasonable amount of ... content".
Definitely interesting as a thought experiment though.
This Mastodon instance is for people interested in technology. Discussions aren't limited to technology, because tech folks shouldn't be limited to technology either!