Lightnews — Scholar-powered news

gbl08ma

@gbl08ma.com

160 followers 360 following 560 posts

Software developer. Most comfortable with Go, can do frontend too. I built jungletv.live, underlx.com, tny.im and more. I know more about the internals of GTA V and Watch Dogs than their average players. Sometimes I make music and high effort shitposts.

Posts Media Videos Starter Packs

Pinned

gbl08ma @gbl08ma.com · Nov 14

"Invalid handle" is Bluesky jargon; it's also what most would like to say when trying to read my username. Too-young-me chose it 15 years ago, and it stuck.

I'm gbl08ma everywhere on the internet, but you can call me just G-B-L or Gabriel. He/him. Feel free to reach out, I think I don't bite.

gbl08ma @gbl08ma.com · 11h

Smells like rage bait [the things artists do these days to get noticed...]

gbl08ma @gbl08ma.com · 11h

It's probably possible to put a SQLite database in a referenced blob...

1 2

gbl08ma @gbl08ma.com · 12h

Put the liquid DnB in a freezer for eight hours so it hardens up.

1 1

gbl08ma @gbl08ma.com · 13h

Why not, I guess. If you feel identified in your understanding of what a "global performer" might be, why not? Who's the title police in this field, anyway?

Reposted by gbl08ma

Shirobon @shirobon.bsky.social · 16h

Really disappointing to see Sabrepulse using AI to create their latest albums.

His statement says otherwise.. but I typed "Math Rock Chiptune" into Suno AI and the results sound the same to me.. especially the drums and leads sound identical.

10 22 84

gbl08ma @gbl08ma.com · 15h

phishing sites at the end of the redirects are in languages you don't understand and impersonate services you never heard of, so it can be tricky for you or for "a magical AI" to realize you're even looking at a phishing page.

gbl08ma @gbl08ma.com · 15h

In my case it was slightly easier to spot the bad ones, because my service would resolve pages to fetch their titles and sometimes I'd see something like my-amazing.domain/awesome-recipe with a "Verizon: Login" title lurking in the database, which is obviously not right. But sometimes the

1 1

gbl08ma @gbl08ma.com · 15h

week. These were good old @gmail.com addresses, too. The links themselves may look innocuous so it's not even something you can throw an LLM or simple statistical model at, you'd need to resolve the links and detect fishy things.

1 2

gbl08ma @gbl08ma.com · 15h

Also, if you think that requiring a Bluesky account will save you, well, maybe it will, but only as long as Bluesky and your service are semi-obscure. In the last years of my shortener, I had restricted link creation to signed-in users and still had to terminate accounts and remove links every other

2 3

gbl08ma @gbl08ma.com · 15h

(at least they never did for me) but they will never stop forwarding things to the "correct" parties either. If you value your domains and servers, I would stop operating the redirection aspect on short.willdot.xyz ASAP.

1 2

gbl08ma @gbl08ma.com · 15h

would be bombarded with abuse reports. And no, Netcraft et al. will refuse exclusively send abuse reports to your designated email, they will always send reports to your infrastructure providers simultaneously. If you're behind Cloudflare, that won't save you: they won't terminate services 6/

2 5

gbl08ma @gbl08ma.com · 15h

the sheer volume of abuse reports will cause more manual work for them than they are usually willing to take (unless you're paying enterprise amounts of money). Hosting a URL shortener is now something I'd only be willing to do under my own IP blocks, mainly to reduce the number of entities that 5/

1 4

gbl08ma @gbl08ma.com · 15h

services, your domain will be taken away from you etc. because these parties won't necessarily do their due diligence; generally speaking, they'll find you to be an annoying customer because even if every abuse report is dealt with by the time they go check (_if_ they bother to go check, that is) 4/

1 4

gbl08ma @gbl08ma.com · 15h

server provider and domain registrar, and even if you process every abuse report automatically (i.e., if you decide to assume every one of these reports is in good faith, and thus not worth investigating manually before proceeding to disabling the links) your server provider will terminate 3/

1 4

gbl08ma @gbl08ma.com · 15h

their own links. If you don't do this, people _will_ create malicious links and you will be drowning in moderation work. Netcraft and similar companies will send emails not just to your designated abuse report contacts (or they'll just guess, abuse@your-domain usually works) but also to your 2/

1 4

gbl08ma @gbl08ma.com · 15h

I ran a URL shortener for over ten years - you can read the third link in my bio to know how that went...

My main word of advice would be: you definitely don't want to operate the redirection service under your own domains and servers. At least, make it so that people MUST use their own domain 1/

1 7

Reposted by gbl08ma

c0ast3r 🇺🇸 @c0ast3r.bsky.social · 1d

Maybe we should start educating school kids at younger age about data management and privacy and the nature of the internet.

Doing so might lead to better outcomes for everyone.

3 16

gbl08ma @gbl08ma.com · 1d

I believe studies have been made showing that the character limits (not just of microblogging platforms, but also of others that hide anything longer behind a "See more..." type toggle) are a contributing factor to the polarization of discourse. However, I don't have the sources right now and I don'

1 5

gbl08ma @gbl08ma.com · 1d

Rice cookers, air fryers and microwave ovens: the trifecta Big Stove doesn't want you to know about.

1 1

Reposted by gbl08ma

Bossett @bossett.social · 1d

wait so the conspiracy is that content moderation is a play to train AI in a sneaky illegal fashion, taking an absurd risk in storing pre-moderation originals, and feeding them into a genAI model... for the purposes of capturing image data before it's posted unprotected to a public social network?

9 28 180

gbl08ma @gbl08ma.com · 1d

Some people are too impatient to accept "moderately better than the alternatives and possibly getting worse for the time being" in silence, but then again, with how everything seems to get enshittified these days, I can't blame them.

gbl08ma @gbl08ma.com · 1d

doutora bluscaia 😂

Reposted by gbl08ma

natalie @natalie.sh · 1d

i would rather ml categorise my images than risk dirtying my eyes to possibly the worst images imaginable. csam and ndii and even worse images if that's even possible??

🍅🥔🫐🌽 hoopy frood 🌶️ 🥑🍫🌵 @huwupy.kawaii.social · 1d

This is one of my favorite takes. I’m sure people would love a 30-day delay on having their images posted so they can be reviewed by hand and also it’s Good Actually to make humans look at some of the most heinous images in existence all day

The Bluesky Head of Trust & Safety admitting that all posted images are processed through an Al autolabeller that presumably employees have no access to or control over? (And probably steals every image to be used in an Al "art" generator, who knows?)
Seriously, this is outrageously irresponsible.

October 7, 2025 at 10:09 AM
20 reposts
1 quote
66 likes
1 save

9 8 97

gbl08ma @gbl08ma.com · 1d

www.youtube.com/watch?v=_zfN...

(Maybe there's a word in the title of this video that's too offensive for Bluesky's ever-changing community guidelines 😒)

AI Slop Is Killing Our Channel

YouTube video by Kurzgesagt – In a Nutshell

www.youtube.com

gbl08ma @gbl08ma.com · 2d

bsky.app/profile/why....

Why @why.bsky.team · 3d

We've been working on a moderation inbox protocol but it hasnt quite shipped yet, it's being prioritized.