PwnallthethingsLots of folks warning that overreliance on AIs can lead to bias.
But that can sound a bit abstract, so let's just leave these examples here.
Kalani Helekunihi
- 36 Followers
- 37 Following
- 16 Posts
I am a time wizard, that can see into the near past.
@ivory I would really like to give this a shot when ready. I am low vision and typically rely on VoiceOver, VoiceControl, and Braille. Tweetbot has been a great experience for years and have been happy to use it!
I would really appreciate having a Mastodon client that works well with accessibility tools - such as not displaying "scrollable area" as is the case with the Electron wrapped app. Hope I can give some useful feedback.
Start simple, prove out the concept, and only after you know you’re on to something should you start tackling the problem in a way that makes sense.
Make tools serve your goals rather than cater your goals to a tool’s limitations.
@jerry Changing profile picture was near instant, whereas a few days back when initially creating my account it was around 20 seconds. You definitely have made some measurable improvements! Thank you.
@racheltobac What I don't understand is why Eli Lilly genuinely thought their official, corporate Twitter account should be named: LillyPad
No wonder journalists were so confused by: EliLillyandCo
Here's a tip... if you want to exfiltrate small chunks of data from an adversarial network - even one that does aggressive content filtering or even a captive portal... DNS Queries. Have a domain somewhere that you can capture request logs, and wait while info funnels in. After all, even a failed DNS Query will still hit your server. And, who doesn't allow DNS lookups? Maybe block some domains for content filtering, but whitelist only trusted ones?
@doot waking up in the morning to see my “unread tweets” under 200 is a pretty clear indication that people are taking a break while basking in the warmth of the dumpster fire.
#AccessibilityTip : Content Warnings
There are some posts here on Mastodon that get hidden behind a Content Warning “click on to display” screen. However… with a screen reader or braille terminal, those just appear as a blank message, offering no indication that there is anything to see.
While it would be wonderful if this gets fixed to at least tell people what the content warning is, or that it’s displayed, there is a setting that simply won’t hide those messages so you don’t need to blind click to see if you’re missing something.
There is a setting available in the WebUI under Settings -> Appearance -> Sensitive Content that will allow “Always expand posts marked with content warnings.”
You can still keep Media (images and video) blurred if you’re worried about not seeing NSFW content.
@jerry that’s really not bad… under 1MB / user / day of data creation. Especially where users are allowed to upload images! I would have expected more.
Thanks for sharing your experiences running this server. It’s fascinating to see the sorts of problems one runs into while scaling up - especially when it’s one guy dealing with it! Even when I’ve overseen growth at work, it’s always been stuff spread around a team so it never felt “real” in the way that your posts have.