man anxiously, desperately trying to sell snake oil insists that the people who refuse to drink snake oil are the real weirdos businessinsider.com/github-ceo…

Working hard between the Website WG and the T&S R&D WG to get out a room upgrading guide for #matrix - we will see how close we get to finish it in time for next week's room version security update release, but it already has lots of useful info in bullet point form: github.com/matrix-org/matrix.o…
If you have the time to help turn it into nice text (or even pictures?!!), please send patches!
in reply to daniel:// stenberg://

My mental model is that curl is a flexible tool for making network calls (of whatever sort you like) whereas wget is a dedicated for downloading and saving webpages or files.

That being said, I usually use curl for the latter use case too, out of muscle memory. But when I spent a month and a half downloading RedHat ISOs over a 28k dial up connection as a kid, you better believe I used wget and it's resume functionality.

Wait, did Movim just unlocked the global search feature in its chat messages? 😱

Thanks to the powerful tsvector/tsquery #PostgreSQL feature (feature not available on MariaDB sorry 😔) you can now search messages by keywords in your history and directly load the conversation from the past 🕰️ !

Just click on the search button or hit Ctrl + F to open the brand new search panel 🔍

The next version will be packed with plenty of awesome features for sure 🚀✨

#xmpp #movim #search #feature

This entry was edited (4 months ago)

Well here's a weird one...

Key sections from the U.S. Constitution have gone missing from the U.S. government's official website, including sections 8 through 10, which include legal provisions like habeas corpus (aka due process). The White House hasn't commented.

techcrunch.com/2025/08/06/key-…

It's here! The NVDA 2025.2 Release Candidate is available: nvaccess.org/post/nvda-2025-2r…

We encourage ALL users to try this before we declare it is stable. Please see tips on how to test without affecting your installed copy in this week's In-Process: nvaccess.org/post/in-process-4…

NVDA 2025.2 includes improvements in Windows 11, browse mode & Microsoft Word. Plus updates to eSpeak-NG & LibLouis, new device support & more!

#NVDA #NVDAsr #Release #Update #NewVersion #PreRelease #test #Testing #FOSS #FLOSS

5 months ago today I was laid off from my job and I am currently looking for my next senior front end developer role.

I have 19+ years of experience working with small agencies and Fortune 500 companies and focus on writing clean, accessible HTML, CSS, and JavaScript.

Resume: aubreysambor.com/resume

Leads/boosts appreciated!

#getfedihired #FediHire

This entry was edited (4 months ago)
in reply to Alexis

@lexipic Thanks!

Your blue screen experience highlights one of the problems with AI. If you show a sighted person a blue screen, they will recognise it & look for the error code & details which might help diagnose the issue. AI will look at it, and get as far as recognising that it looks similar to images it has seen previously of blue screens.... but not understand what the text means or which part is importnat in that. It will see the large :( as just as important and describe that to you.

in reply to NV Access

I did get sided help, and it was eventually discovered that the blue screen was error code for a hardware failure. The computer could not be saved. I did however, managed to get some files off of it that I wanted back. So I’m not overly upset about that. However, I am upset about all the time lost. I had to go searching for a charger for my latitude, and that took a whole week to find one. So I was a week without a computer, and was left not knowing what I was going to do about school once it starts for me in December. It was pretty bad. You can have all the software you want, yet, when the hardware fails, that’s where it becomes devastating.

Some postmortem notes about our 19th mapping party…

en.osm.town/@OSMIndia/11493311…

On a positive note, I think this was a productive party. 8 out of 10 participants were experienced mappers, and I’m happy with the data we added to the market. 👍

On the other hand, there were hardly any people from Noida - clearly, the publicity needs more work 😐

One participant recommended posting on Luma, so I guess we’ll be trying that.

Also, if you like my work, please support me on Liberapay. I’m unemployed, and rely on the community’s support for my contributions.

#OpenStreetMap #OSM #contraMaps

This entry was edited (4 months ago)
in reply to Traveling OSM Salesman^W Problem ✊🏳️‍🌈🏳️‍⚧️🇺🇦

Even on the night before the party, we were agonizing over the perpetual question - what editor to teach to newcomers?

And after the workshop and the latest mapping party, I think I have a new and improved idea.

  1. We always introduce #CoMaps (and earlier Organic Maps) to newcomers, focusing on it as a means of using OSM. But we avoid using its editor - offline data can be confusing and (in the presence of concurrent editors) can more easily lead to duplication.

But if we want people to recommend CoMaps to other, less technical people…they may want the latter to be able to add places without installing another app.

  1. I have long admired StreetComplete’s gamified interface, but always figured that it would cause issues in a mapping party.

But perhaps we can incorporate it after all? The people it is most likely to cause issues for are the #Vespucci users, and they are experienced enough (and Vespucci’s tooling is good enough) to handle upload conflicts.

Another issue is that if a user adds some data using a #StreetComplete layer, it takes time for quests to appear for it.

  1. #MapComplete is good because it puts together the things people coming from Google Maps often expect from OSM - map data, photos, and ratings+reviews. The questions and the photo-taking may also add some variety and fun, and the photos also help other mappers verify the data and add details.

So here’s the new idea -

  1. We practice adding 1-3 POIs with CoMaps. There may be duplication, but that’s okay, we (the organizers) will clean it up later as we usually do.
  2. Then, we add 5-10 POIs with MapComplete.
  3. Finally, we try out StreetComplete to add more details to the streets and POIs. We will use SC’s team mode to divide quests between people on the same team.

I’ll test the viability of this process soon.

(Vespucci will go back to being an editor for regular participants.)

#OpenStreetMap #OSM #contraMaps

This entry was edited (4 months ago)
in reply to Andreas Wiedenhoff

If it is still around, here is a map of old railway gear. Onoy requirement to be added is to be permanently in the same place: mapcomplete.org/historic_rolli…

#OpenStreetMap #MapComplete

🎨 Design the future look of the euro!

The European Central Bank has launched a design contest for the new euro banknotes.

Open to graphic designers in the EU, submit your proposal for one of two themes:
🖼️ European culture
🌍 Rivers and birds

The designs – reflecting Europe’s shared cultural identity and natural heritage – will shape our currency for years to come!

🗓️ Apply by 18 August 2025, 12:00 CET

🔗 More info: europa.eu/!cWfmdT

This entry was edited (4 months ago)

Grr! For some reason Windows decided to turn on defragmentation for an SSD. Who does that? I turned that shit off! In other news, want to know stuff about your SSD like temperature, how many gigs it's transfared over the years, how many hours it's been powered on? Check out the free utility Crystal Disk info: crystalmark.info/en/software/c…

Zajímá vás #politika, ale ne zase tak moc, abyste denně sledovali vývoj?

#Volby jsou za rohem a tak by se vám mohlo líbit měsíční shrnutí dění od @Programy do voleb - nyní i na #VHSky - hezky pěkně jak to máme rádi bez trackerů, reklam a svobodně :)

A když si vzpomenete, nezapomeňte, že je tam tlačítko "podpořit"

vhsky.cz/c/programy/videos

Update: This piece is getting some interesting pushback from parents who think I'm being alarmist about AI toys over on my other social platforms.

On the other side of that, I'm hearing so many people taking the usual "AI BAAAAD" stance, some of those people thinking I agree with them simply because I took a hardliner stance in this post.

For clarification: I'm not anti-AI. I use these tools daily for my research and writing as well as accessibility aids to offset some of the disadvantages I face due to my blindness. I study AI from the computer scientist perspective and am studying to be an elementary teacher precisely because I see AI's educational potential. I'm not even entirely against the idea of AI companionship, if it's framed right.

What actually bothers me is the business model. When Moxie robots suddenly "died" last year because the company went under, kids had to grieve their artificial friend. Parents got a scripted letter to explain why their $799 companion stopped talking which provided little comfort to kids who experienced digital abandonment. Trust me, the videos I've seen of kids crying because their beloved friend unexpectedly died over night is truly heartbreaking.

That's no glitch, that's what happens when you outsource childhood relationships to venture capital that only cares about investment returns.

The real question isn't whether AI toys are inherently bad. It's whether we're okay with corporations experimenting on our kids' emotional development while claiming it's "age-appropriate play."

What are your thoughts? Let me know in the comments.

open.substack.com/pub/kaylielf…

#AIToys #ChildPrivacy #ChildDevelopment #DigitalRights #TechEthics #SurveillanceCapitalism #COPPA #DataPrivacy #ChildSafety #TechRegulation #DigitalLiteracy #ParentingInTheDigitalAge #EdTech #CorporateAccountability #TechCriticism #EthicalTech

#AndroidAppRain at apt.izzysoft.de/fdroid today brings you 11 updated and 2 added apps:

* Compete: a place where people can improve themselves in a positive way while having a lot of fun 🛡️
* Stario: a minimalist, decluttered launcher 🛡️

RB status: 678 apps (51.2%)

Enjoy your #free #Android #apps with the #IzzyOnDroid repo :awesome:

"This is how Israel killed this hungry child.
He walked through bombs and hunger to reach Zikim, hoping to bring flour back to his family...
But an Israeli bullet got there first.
He returned carried on wood, wrapped in the same flour bags he went to get.
Israel kills the starving every day — and the world stays silent.
Save us."

x.com/AnasAlSharif0/status/195…

#Palestine #Gaza #Israel #GazaIsStarving
#AnasAlSharif
@palestine

Let's dive into the Wednesday teaser edition of Other People's Music! This thread offers the first line of each of the week's reviews, plus links to the music, socials and full review. If you prefer to dive into the full version, you can find it here:

etherdiver.com/2025/08/01/opm-…

#Music #MusicDiscovery #OtherPeoplesMusic #OPM #Musodon

Article 1, Section 9 and 10 have been removed from the official government website on the US Constitution. This better be some kind of joke.

Among numerous points:
•Section 9 forbids suspending The Privilege of the Writ of Habeas Corpus
•Section 10 forbids ex post facto laws

Removing these two would absolutely mean full blown dictatorship. Genuinely hoping this is some sort of prank or joke, but you never know with this fascist regime.

You may have noticed billboards for something called "PDX Crusade" popping up in recent weeks. This event is being sponsored by Athey Creek aka Proud Boy Church, since a number of Oregon Proud Boys and their associates attend services there.

Portlanders should be aware that a number of out-of-town far-right chuds may be roving around before & after the event happening this Sat & Sun - August 2 & 3 - at the Moda Center. Please take appropriate precautions.

The Christian concept of "crusading" has a long and bloody history of hate, and the refrain from the far-right claiming they need to "save" or "cleanse" a city that isn't theirs is one we've heard in Portland many times in recent years.

We see this christofascism writ large in the actions of the current administration but it's especially visible in Defense Secretary Pete Hegseth who has tattoos of a crusader cross, and the words "deus vult" ("god wills it") and "infidel".

For more information on Athey Creek, check out this thread:
bsky.app/profile/sometimespdx.…
For more information on christian nationalism/christofascism, check out these articles:
projects.propublica.org/christ…
philpapers.org/archive/FOEASH.…

The Trump administration is warning states that they risk losing access to broadband funds if they set price caps on low-cost ISP plans (Jon Brodkin/Ars Technica)

arstechnica.com/tech-policy/20…
techmeme.com/250805/p50#a25080…

#Starlink is deliberately de-orbiting older satellites to ‘burn up’ in the pristine upper atmosphere.

‘Burn up’ actually means ‘leave a bunch of metal vapour in the upper atmosphere’.

—————
quote

Before the first Starlink launches began in 2019, only about 40 to 50 satellites re-entered per year. SpaceX just brought down ten years' worth in only six months, adding an estimated 15,000 kilograms of aluminum oxide to the upper atmosphere.
—————

Scientists are only just studying what this may do to the atmosphere (and life on earth). Early indications are not good.

People should be *really* worried about this and keep an eye out for future updates. Especially given the #USA seems to be turning into a lawless joke.

Cc @sundogplanets

#Satellite #Pollution #Musk

spaceweather.com/archive.php?v…

in reply to Devorppa

> SpaceX just brought down ten years' worth in only six months, adding an estimated 15,000 kilograms of aluminum oxide to the upper atmosphere.

20 seconds of research on this:

>> meteors from space deposit 100 to 200 tons of metallic material every day across the globe
>> The meteors contain Tri-methyl aluminum, Lithium, Barium

why is 15,000kg of satellites a bigger concern than 100-200 tons of meteors (PER DAY!!)?

This entry was edited (4 months ago)
in reply to Prof. Sam Lawler

also, what exactly was our plan if we just started getting bombarded with a lot more meteoroids?

We're just going to have to get used to the idea that we need to start modifying the upper atmosphere.

We're already screwed because we have a significant lack of cloud coverage due to the existing warming and it's getting worse. So we're going to need to intervene anyway. We are going to have to increase the albedo ourselves and start doing things like stratospheric aerosol injection

in reply to Jan Niklas Fingerle

an assertion was made. A curious mind would say "well how much is that in relation to everything else naturally entering the atmosphere?". So I found that answer, I posed a question, and I received a useful answer.

If you're trying to make some cutesy "stop being a smartass, just trust the scientists" point you can eat a bag of dicks. The scientists are wrong a *lot*.

Example: the climate models are all wrong. And not in a good way. Why? Compounding floating point errors. It's far worse than they're projecting.

Another reason: the calculations have not taken into consideration the lack of cloud coverage caused by the current amount of warming. How did they miss this? @SlicerDicer introduced this problem to me probably ... 8 years ago now? Maybe longer? Aren't the scientists supposed to be good at science? Oh wait, the lack of interdisciplinary expertise caused them to completely ignore the gas physics that rebreather diving experts know. Turns out that the warmer atmosphere affects the ability for clouds to condense. But nobody baked it into any climate models.

science.org/content/article/ea…

in reply to feld

Yep, the gas physics are clear. It was about 10 years ago.

Why would they bake it into the climate models the entire thing was that more heat = more vapor. Superficial understanding says more clouds. Till you run the math of what’s actually happening to layers of the atmosphere. Then it’s very clear.

Dunno, it’s like maybe if their life depended upon the math being right and correct they’d take it serious?

Can’t say it’s incorrect when my life depended on it being 100% correct. 7 mins to dead if you are wrong.

Seven. Minutes. To. Dead.

That’s assuming everything is perfect when things go wrong and you notice it.

@pesh you may be interested in this!
Be My Eyes and Amtrak Partner to Pilot Innovative Visual Interpretation Technology for Blind and Low Vision Passengers - Be My Eyes for Business bemyeyes.com/business/news/be-…

I needed to setup a rather annoying test case for Pleroma. It would have taken me 20-30 mins to get it all sorted. This "AI" garbage did it for me in 2 minutes. I provided it two prompts:

First prompt:

---
Please analyze and document for me in this chat the behavior of Pleroma.User.Search.search/2 and how it handles searching for user accounts, the way the Postgres ranking works, etc
---

(it produced a bunch of documentation and code citations that matched reality)

Second prompt:

---
ok, this basically matches what I'm seeing as well. However, there appears to be a bug somewhere. Let's begin by creating a test directly for Pleroma.User.Search.search/2 that demonstrates what I am seeing.

First, let's create these users:

- one user for ourself
- lain@cyber.ms
- Laine@pialle.fr
- lain@lain.com
- lain@pleroma.soykaf.com
- lain@fediffusion.art

Second, let's establish relationships

- we have no relationship with lain@cyber.ms
- we have no relationship with Laine@pialle.fr
- there is mutual following relationship for lain@lain.com
- there is mutual following relationship for lain@pleroma.soykaf.com
- we are following lain@fediffusion.art

Now, when searching for "@lain" I expect the results to be in this order:

- lain@cyber.ms
- Laine@pialle.fr
- lain@lain.com
- lain@pleroma.soykaf.com
- lain@fediffusion.art

Please write a test demonstrating this exact behavior, and then we can proceed with altering the behavior as a follow-up.
---

It wrote the tests. With better details than I would have. The test case produced this output:

=== Search Results Debug ===
1. lain@lain.com (rank: 2.333333373069763, id: AwtmDkvx8hzBOktYjg)
2. lain@pleroma.soykaf.com (rank: 1.8125000596046448, id: AwtmDkwf64YLQxE7qC)
3. lain@fediffusion.art (rank: 1.609523868560791, id: AwtmDkx14kpvS3OPOS)
4. lain@cyber.ms (rank: 1.3571429252624512, id: AwtmDkvFBLQ1MYYzdA)
5. Laine@pialle.fr (rank: 0.8067227005958557, id: AwtmDkvbA1hbNejHBQ)

and then it thought that I wanted the actual broken behavior, haha, because I expected the test to pass and match the broken behavior I described:

---
The test is now working and clearly shows the current behavior vs. the expected behavior. Let me analyze the results:

Current Actual Order:
lain@lain.com (rank: 2.33, mutual following)
lain@pleroma.soykaf.com (rank: 1.81, mutual following)
lain@fediffusion.art (rank: 1.61, following)
lain@cyber.ms (rank: 1.36, no relationship)
Laine@pialle.fr (rank: 0.81, no relationship)

Expected Order:
lain@cyber.ms (no relationship, exact match)
Laine@pialle.fr (no relationship, similar match)
lain@lain.com (mutual following)
lain@pleroma.soykaf.com (mutual following)
lain@fediffusion.art (following)

The issue is clear: the social relationship boosting is overpowering the base trigram similarity scores, causing users with relationships to always rank higher than users without relationships, regardless of how good the text match is.

Looking at the ranking scores:
Users with mutual following get a 1.5x boost
Users with one-way following get a 1.3x or 1.1x boost
Users without relationships get no boost

This means even a poor text match with a relationship will rank higher than a perfect text match without a relationship.

The test successfully demonstrates the bug you mentioned! The current implementation prioritizes social relationships over text match quality, but your expected behavior prioritizes text match quality over social relationships.

Would you like me to proceed with modifying the search algorithm to match your expected behavior, or would you like to discuss the approach first?
---

Please tell me again how this technology is terrible and nobody should use it? It just saved me an hour
RT: friedcheese.us/objects/0105bfc…


ooookay, wow, what the hell is going on with this?

the attached image is the autocomplete for "@lain" that I get. This is the order of results. To clarify:

1. lain@cyber.ms (no relationship, exact match)
2. Laine@pialle.fr (no relationship, similar match)
3. lain@lain.com (mutual following, exact match)
4. lain@pleroma.soykaf.com (mutual following, exact match)
5. lain@fediffusion.art (following, exact match)

I recreated this in a test case and ran it expecting these results. Except I get different results:

=== Search Results Debug ===
1. lain@lain.com (rank: 2.333333373069763, id: AwtmDkvx8hzBOktYjg)
2. lain@pleroma.soykaf.com (rank: 1.8125000596046448, id: AwtmDkwf64YLQxE7qC)
3. lain@fediffusion.art (rank: 1.609523868560791, id: AwtmDkx14kpvS3OPOS)
4. lain@cyber.ms (rank: 1.3571429252624512, id: AwtmDkvFBLQ1MYYzdA)
5. Laine@pialle.fr (rank: 0.8067227005958557, id: AwtmDkvbA1hbNejHBQ)

Literally the exact order that I want!

Our code is doing the right thing. So why is it not working in practice?