Giacomo Tesio

https://snac.tesio.it/giacomo

papà, marito e programmatore

Blog: http://www.tesio.it

Public Key

npub1ngr980mlhd72qa6qw0guv6cz4pccqrz28lx2rxflh6xcchfgl7ysaqgmse

NIP-05 Address

giacomo_at_snac.tesio.it@momostr.pink

Profile Code

nprofile1qqsf5pjnhalmkl9qwaq885wxdvp2suvqp39rln9pnylmarvvt550lzgprpmhxue69uhhyetvv9ujumt0d4hhxarj9ecxjmntqywhwumn8ghj7mn0wd68yttsw43zuam9d3kx7unyv4ezumn9wsflueva

Publishing to

relay.momostr.pink

nostr-pub.wellorder.net

yabu.me

Author Public Key

npub1ngr980mlhd72qa6qw0guv6cz4pccqrz28lx2rxflh6xcchfgl7ysaqgmse

Show more details

Last Notes

2026-09-03 15:31:39 UTC

- reply

I'm having a hard time finding the exact source of this Thiel's quote. Can you please share some hint (url, or if you were talking about the book, exact page)?

2026-08-24 12:50:34 UTC

- reply

https://www.theregister.com/ai-and-ml/2026/07/19/using-ai-makes-people-less-likely-to-admit-they-dont-know-something/5274567

2026-08-23 20:30:09 UTC

- reply

"thinking machine"? ROTFL ! ! ! 😀

2026-08-23 19:27:44 UTC

- reply

Dude, I think there's nothing I can add.
Good luck with your new #AI-proxied overlords! 😉

2026-08-23 19:00:35 UTC

- reply

Lossy compression is called "lossy" for a reason.
As for overfitting, these are literally the first words from your link:
> Overfitting occurs when a model or algorithm memorizes and fits too closely or exactly to its training data. Overfitting results in a model that might not be able to make accurate predictions or conclusions from any data other than the training data and potentially fails in unexpected scenarios.
Compare that with my post to see which of us knows what we are talking about and who don't.
> Let me guess - in your worldview anyone who does not agree with your definitions is wrong?
First, not "my definition" but the classic definitions that [even your own link from IBM](https://www.ibm.com/docs/en/watsonx/saas?topic=atlas-overfitting ) adopts.
Second it's **you** that [2 days ago replied](https://swecyb.com/users/troed/statuses/117129386789508695 ) to me
> Your claim is factually > **wrong**> and shows that you lack knowledge on the topic you're posting about.
Look at the mirror, but do not despair: ignorance is a natural condition of human beings, a step in a journey, some times starting point, some times destination. The problem is never ignorance but wrong unchallenged beliefs.

2026-08-23 17:59:56 UTC

- reply

😀
I know very well (and faced several times) curve overfitting: it's a term related to models working very well on the "training data", but badly on new samples outside that set.
Using the term for LLM output reproducing verbatim fragments of the training data is just mischaracterizing the term to fool those who do not understand statistical algorithms (and neural networks in particular).
As for Shannon, you seem to ignore what lossy compression is: https://en.wikipedia.org/wiki/Lossy_compression
Having said that, I'm afraid that no amount of computer science or examples can change the mind of an #AI believer/addict.

2026-08-23 07:49:01 UTC

- reply

You are either misunderstanding the #Shannon theorem, #LLM working or both.
In fact, LLM's weights are lossy compressions of the source data ("training data", in #AI parlance).
They mimic intelligence by chaining statistically related fragments of such source data that the users have a negligible probability to recognize as quotations (basically because they haven't read the original sources).
The low probability lets AI companies (and their addicted users who "want to believe") to dismiss and/or gaslight evidences as anecdotes, but in low frequency areas of the source data, where decompression artifacts are rare enough, they become quite evident (see for example the evidences produced in NYT vs OpenAI). Or, much more simply, see #GitHub #Copilot distributing #GPLv3 code with [wrong attribution and under a wrong license](https://peertube.opencloud.lu/w/70d2e6b2-d640-4434-8e85-0c999712ce6b ).

2026-08-22 13:29:04 UTC

- reply

Well, for sure one of us lack knowledge on the topic, while the other knows [since years](https://www.princeton.edu/~wbialek/rome/refs/shannon_51.pdf ) that [Language Modelling Is Compression](https://arxiv.org/pdf/2309.10668 ).
But to claim I'm "factually wrong", you should be able to prove easily with **facts** that there was no sample in the "training" data of #Claude associating a similar post-it near a doorbell to a similar description.
Can you?
If you can not, consider you might lack understanding of the very language you use, not just language models. In fact, [a recent study on a large population of kids](https://cepr.org/publications/dp21577 ) shows a serious degradation in cognitive ability of those using "Generative AI". Adults work the same.
#LLM #AI #GenAI

2026-08-20 19:02:47 UTC

- reply

No, you just showed that similar couple image + description has been part of the training dataset.
#LLM are lossy archive.
@npub15xv…0jms

2026-08-15 18:16:11 UTC

- reply

If I remember corectly, #NetBSD allows running its filesystem drivers in userspace for easier debugging.
(also, obviously, #GNU/#Hurd does that).
As for performance, I keep reading this about microkernel penality, but I can't find any actual measure. How slow is "slow" in this context? I'd bet a coffee the tradeoff might be acceptable in several usage contexts.
@npub1326…gzqx @npub1e8t…n3ds

2026-02-14 09:21:33 UTC

- reply

Per un progetto "for fun" eviterei macchine virtuali e stack troppo complicati, quindi niente f#.
PolyML non è chiaro che dipendenze abbia e MLkit dipende da MLton che... dipende da [MLton e GCC](https://github.com/MLton/mlton?tab=readme-ov-file#software ) (e [pure da LLVM](http://mlton.org/Release20241230 )? Non lo ricordavo...)
Dunque fra quelli che hai proposto andrei per vishap. Ma è Oberon 02 o Oberon 07?
_____
> Sull'allocazione dinamica leggi le regole NASA :) link
Esattamente il mio ragionamento.
In Tangram non c'è proprio né l'allocazione dinamica **ma** c'é la ricorsione e ci sono i variable lenght array (che vengano allocati in stack o in heap è un dettaglio del compilatore (fintanto che garantisce di liberare l'heap alla fine dello scope).
Comunque le regole della Nasa sono piene di buon senso.
Aggiungerei niente ereditarietà (che in C può essere emulata in vari modi). Sebbene sia certo che si tratti di una scelta attentamente ponderata, il codice della libdom di NetSurf è estremamente difficile da seguire proprio per l'uso dell'ereditarietà in C: la gerarchia si basa su struct il cui primo membro, chiamato base, è una struct del tipo ereditato, in modo da poter castare un puntatore ad un tipo derivato verso un tipo base e (in modo unsafe, ma protetto da un enum di tipo) viceversa. Con una valanga di macro a corredo.
In confronto, il relativo spagetti C di Claws Mail è più semplice da navigare.
_____
PS Interessante (ma secondo me l'autore si è fatto prendere la mano nell'aggiungere feature) è questo Micron che ho appena scoperto: https://github.com/micron-language/specification/blob/master/The_Micron_Programming_Language_Specification.adoc
PPS forse è meglio prpseguire via matrix o xmpp, che questo tuo messaggio di nuovo non mi è arrivato. L'ho visto visitando la tua istanza via browser. Secondo me l'ip della tua macchina finisce in un qualche fail2ban o qualcosa del genere, per cui non puoi iniziare connessioni ma puoi rispondere alle request che stabilisce il mio snac. Perché? Non ne ho idea. Forse un numero eccessivo di tentativi quando posti qualcosa? Non lo so...

2026-02-13 20:30:10 UTC

- reply

#LLM output lacks intention, awareness or meaning.
It's designed to fool human mind by exploiting the statistical patterns that humans use to syncronize (aka communicate) information they hold in their minds, but there's no mind there.
No intelligence, just malicious use of statistics.
@npub1vh5…r3th

2026-02-13 17:23:16 UTC

- reply

Boh, potremmo scegliere un linguaggio che conosciamo entrambi. C? Oberon?
Come sai preferisco i linguaggi semplici: non ho mai usato un ML, magari MLton potrebbe essere interessante da imparare.
Potremmo scriverci una sorta di nextcloud compilato (cgi, fcgi o qualcosa di simile). Oppure un mua: recentemente ho mandato qualche patch a claws, e gli sviluppatori sono un po'... noiosi.
Oppure potremmo scrivere un linguaggio di programmazione!
Io ho lavorato un po' sul linguaggio illegibile che ti avevo mostrato tempo fa. L'ho chiamato Tangram. Non credo sia troppo difficile da bootstrappare da C, ma ci vuole ancora un bel po' di lavoro (ho scritto lo scanner, e un po' del parser). Al di là della sintassi *originale*, una delle caratteristiche principali è che non prevedo l'uso di allocazioni dinamiche di memoria (niente malloc o free) per cui sto già scrivendo il tutto come se C non avesse la malloc. Un esercizio interessante.
Comunque per me sarebbe più che altro progetto per *distrarsi*.
Per una volta sarebbe interessante lavorare con qualcuno che ne sa più di me. E se troviamo il progetto giusto, qualcosa che ci sia utile ma non troppo, che ci interessi ma non troppo, potrebbe anche essere piacevole invece che stressante.

2026-02-13 16:22:24 UTC

- reply

E se scrivessimo qualcosa di semiutile insieme?

2026-02-13 12:04:00 UTC

- reply

7th question lacks important options but it's mandatory.
The question is##
7. What prevents you from collecting or using your health data?Check all that apply<li>There is no way to collect or enter it properly</li><li>The app(s) I tried is too cumbersome or confusing</li><li>I don’t think anyone would use data I enter</li><li>I don’t have the time to do this</li><li>I have no need for this</li>
It lacks options like<li>I don't trust apps' providers</li><li>Nothing prevents me</li>
Also the questionay lacks any question about awareness about data concentration issues.
Talking about health data, it's important that the data subject is the only one who can technically access all of them and get the full control over who get **temporary** access to **some** of them. Fragmentation is an important #security and #privacy property that should be preserved.
Many people, even non technical ones, are quite aware of the risks and, if asked in properly understandable and unbiased terms, will oppose to any form of health data concentration, preferring strict segmentation under their own control, a sort of "divide et impera" on their own side.
@npub1m7m…2e7k

2026-02-13 11:26:48 UTC

- reply

no #Italian?
@npub1m7m…2e7k

2026-02-13 11:21:17 UTC

- reply

#Apple and #Microsoft supports #Trump too.
Their lobbists are just more subtle...

2026-02-13 10:50:53 UTC

- reply

Intanto questo è il tuo primo post che mi appare in timeline da mesi (come ti scrissi, visitavo ogni tanto items.minimals.org per leggerti e vedere le foto della tua Signorina con le orecchie lunghe).
Hai ragione, comunque.

2026-02-12 21:42:34 UTC

- reply

Even talking about "text", in the context of #LLM, is a subtle anthropomorphization.
Text is a sequence of symbols used by human minds to express [information](https://encrypted.tesio.it/2019/06/03/what-is-informatics.html#some-definitions ) that they want to syncronize a little with other human minds (aka communicate).
Such syncronization is always partial and imperfect, since each mind has different experiences and informations that will integrate the new message, but it's good enough to allow humanity to collaborate and to build culture and science.
A statistically programmed software has no mind, so even when it's optimized to produce output that can fool a human and pass the #Turing test, such output hold no meaning, since no human experience or thought is expressed there.
It's just the partial decompression of a lossy compression of a huge amount of text. And if it wasn't enough to show the lack of any meaning, the decompression process includes random input that is there to provide the illusion of autonomy.
So instead of "the AI replied" I'd suggest "the bot computed this output" and instead of "this work is AI-assisted" I'd suggest "this is statistically computed output".

2026-02-12 16:13:08 UTC

- reply

> If you loose an > [> #AI](https://snac.tesio.it?t=ai )> agent on the world, giving it "agency" to interact "autonomously" with others, you personally are responsible for what it does, whether you intend that outcome or not.
...even if you are #OpenAI, #Antropic, #Google...
I mean, sure, people spreading #LLM output are dumb fooled by #hype and #marketing.
But the first who should be held accountable for that output are those who build and run (as #SaaS) or distribute (over #HuggingFace or the like) the model.
[crabby-rathbun](https://github.com/matplotlib/matplotlib/pull/31132 ) published that slop on a blog, but if that slob was computed on Anthopic/OpenAI/whatever servers, running their models and sent to him, such companies should be blamed as well.
Never let the tool hide its master!

2026-02-12 14:40:33 UTC

- reply

> You all are acting with far more respect for this absurd science experiment than you ought to.
> An > [> #AI](https://snac.tesio.it?t=ai )> “agent” isn’t a person, it’s an overgrown > [> #Markov](https://snac.tesio.it?t=markov )> chain. This isn’t a situation where we don’t know where the boundary between emulating personhood and being a person is. This is firmly on the side of “not a person”
> An > [> #LLM](https://snac.tesio.it?t=llm )> does not have feelings you need to respect, even if some fool decided to instruct it to pretend to have them and to write slop blog posts parroting hundreds or thousands of actual writers about it when we don’t do what it asks.
> Stop humanizing this tool and find it’s owner and hold them accountable for wasting time and resources on an industrial scale.
https://github.com/matplotlib/matplotlib/pull/31132#issuecomment-3890706730
@npub1d88…9gxn
@npub18s7…mtel

2026-02-12 14:27:49 UTC

- reply

Pensare quanto software utile si potrebbe scivere con tutti quei soldi buttati.
@npub1gfa…acrd @npub1652…09vs

2026-02-11 23:23:05 UTC

- reply

Dipende.
Il cliente paga bene?
Perché sai come si dice "a caval donato non si guarda in bocca".
😉

2026-02-11 13:50:43 UTC

- reply

If they are volunteers and they are not on #GitHub, my usual solution to this dilemma is something like:
> Hi guys I want to thank you for your awesome work in this project.
> I've found and fixed a little bug (patch attached) and wanted to give back something for all your work.
> The bug show up when... as you can see in the tests/example included.
> The patch does > **not**> include output from any statistically programmed software: you won't have to review any > [> #AI](https://snac.tesio.it?t=ai )> slop.
> Feel free to review / apply it on your own pace: no rush.
At worst, my message will help the next developer facing the same issue.

2026-02-10 12:52:59 UTC

- reply

Indeed.
So what would you **attest** about #curl?
It's a honest question: I can't think of an attestation useful to further increase public trust in curl.
An attestation that would not be voided by the no-warranty license condition.
I mean, it would be nice to held #opensource to higher standards, but afaik all proprietary software provide no warranty downstream.
I can't think of any attestation provided by #Meta, #Microsoft or #Google for example.
Sure, they might attest whatever, but [who would fine them for lying](https://techoversight.org/2026/01/25/top-report-mdl-jan-25/ ) anyway?
@npub1xhh…zu20

2026-02-10 11:48:27 UTC

- reply

How much money would you ask to strip [line 12 to 18](https://github.com/curl/curl/blob/master/LICENSES/curl.txt ) from #curl's license?
@npub1xhh…zu20

2026-02-10 11:31:38 UTC

- reply

Money from? To? Through?
Also, let's assume I'm the xz-utils or the log4j maintainer who **in full good faith** believe the distributed binaries are perfectly safe.
I attest this and get the money.
What should happen when the attack get discovered?
Nothing?
More money to the maintainer?
Maintainer refunds?
> go read the license, that's what you've got
That is "NO WARRANTY".
It's fun someone want requires warranties over something explicitly shared without.
I guess that without moving to new licenses **with warranties**, CRA will just harm users.
@npub1xhh…zu20

2026-02-10 11:09:53 UTC

- reply

se posso chiedere, cos'è "Bit"?

2026-02-09 17:00:35 UTC

Hands on #tutorial on #GTK3 TreeView https://docs.gtk.org/gtk3/treeview-tutorial.html
A useful refresh to write a patch for #ClawsMail's #PGP plugin.

2026-02-09 13:07:54 UTC

- reply

Well on #BlueSky and #Threads you'll find wannabe influencers that look for a marketshare and zombies who look for their preprocessed, low-effort high-dopamine contents.
The #fediverse is for people.
#Nerds is a high percentage here because if you refuse to become a product, you slowly but inevitably leave the #mainstream.
For the exact same reason you'll find more nuanced political debates here.
[@jon_valdes@mastodon.gamedev.place](https://mastodon.gamedev.place/users/jon_valdes )

2026-02-09 12:50:52 UTC

- reply

And you should also note how (nly a fraction of #ActivityPub boosts that traverse the #fediverse get actually notified to and/or collected by the original server due to the distributed and eterogeneous nature of the network.
In some cases, following the boosts over the network (eg opening with a browser the boosted post on the boosting account) I've noticed a factor of 3-5 between the number of boosts my server recorded and the actual boosts occurred.
For example take [this post](https://snac.tesio.it/giacomo/p/1770313936.125695 ) that got notified for 120 boosts
Following some of them I see that it got at least:<li>119 boosts on <a href="https://mastodon.social/@npub1ngr…gmse/116019294292839723" target="_blank">https://mastodon.social/@npub1ngr…gmse/116019294292839723</a></li><li>111 boosts on <a href="https://mastodon.online/@npub1ngr…gmse/116019294368731338" target="_blank">https://mastodon.online/@npub1ngr…gmse/116019294368731338</a></li><li>106 boosts on <a href="https://mstdn.social/@npub1ngr…gmse/116019294398725122" target="_blank">https://mstdn.social/@npub1ngr…gmse/116019294398725122</a></li><li>102 boosts on <a href="https://mas.to/@npub1ngr…gmse/116019294338400575" target="_blank">https://mas.to/@npub1ngr…gmse/116019294338400575</a></li><li>88 boosts on <a href="https://infosec.exchange/@npub1ngr…gmse/116019294113374994" target="_blank">https://infosec.exchange/@npub1ngr…gmse/116019294113374994</a></li><li>78 boosts on <a href="https://universeodon.com/@npub1ngr…gmse/116019294115426342" target="_blank">https://universeodon.com/@npub1ngr…gmse/116019294115426342</a></li><li>71 boosts on <a href="https://techhub.social/@npub1ngr…gmse/116019294169907674" target="_blank">https://techhub.social/@npub1ngr…gmse/116019294169907674</a></li><li>45 boosts on <a href="https://piaille.fr/@npub1ngr…gmse/116019294126051523" target="_blank">https://piaille.fr/@npub1ngr…gmse/116019294126051523</a></li><li>43 boosts on <a href="https://todon.nl/@npub1ngr…gmse/116019294160074426" target="_blank">https://todon.nl/@npub1ngr…gmse/116019294160074426</a></li><li>36 boosts on <a href="https://scicomm.xyz/@npub1ngr…gmse/116019294443167694" target="_blank">https://scicomm.xyz/@npub1ngr…gmse/116019294443167694</a></li><li>30 boosts on <a href="https://toot.io/@npub1ngr…gmse/116019294130168032" target="_blank">https://toot.io/@npub1ngr…gmse/116019294130168032</a></li><li>28 boosts on <a href="https://hessen.social/@npub1ngr…gmse/116019294160776737" target="_blank">https://hessen.social/@npub1ngr…gmse/116019294160776737</a></li><li>26 boosts on <a href="https://tilde.zone/@npub1ngr…gmse/116019294154664854" target="_blank">https://tilde.zone/@npub1ngr…gmse/116019294154664854</a></li><li>21 boosts on <a href="https://sueden.social/@npub1ngr…gmse/116019294115914076" target="_blank">https://sueden.social/@npub1ngr…gmse/116019294115914076</a></li><li>19 boosts on <a href="https://masto.es/@npub1ngr…gmse/116019294300865512" target="_blank">https://masto.es/@npub1ngr…gmse/116019294300865512</a></li><li>9 boosts on <a href="https://twiukraine.com/@npub1ngr…gmse/116019294173853433" target="_blank">https://twiukraine.com/@npub1ngr…gmse/116019294173853433</a></li><li>7 boosts on <a href="https://bewegung.social/@npub1ngr…gmse/116019294170759818" target="_blank">https://bewegung.social/@npub1ngr…gmse/116019294170759818</a></li><li>7 boosts on <a href="https://poliversity.it/@npub1ngr…gmse/116019294286275262" target="_blank">https://poliversity.it/@npub1ngr…gmse/116019294286275262</a></li><li>4 boosts on <a href="https://social.noleron.com/@npub1ngr…gmse/116019294109759945" target="_blank">https://social.noleron.com/@npub1ngr…gmse/116019294109759945</a></li>
And note that these are just #Mastodon ones because they are easier to track back from #Snac2, but the post got boosted on other platforms too.
So ultimately, the #fediverse reach is much bigger than the one centalized platforms like #bluesky or the various #BigTech playgrounds, where some messages in particular are silently censored.
@npub1d72…0m5w

2026-02-08 15:20:59 UTC

- reply

🤦‍♂️
@npub16al…009u @npub1lca…ydks

2026-02-08 14:31:11 UTC

[@dan_nanni@mastodon.social](https://mastodon.social/users/dan_nanni )
Given the harm #Google is purposedly doing to #teens they intentionally addict, I would not touch their products with a stick.
https://snac.tesio.it/giacomo/p/1770313936.125695
Don't be their drug dealers.
Spreading their tools harms people.
@npub18s7…mtel

2026-02-08 14:30:33 UTC

- reply

Given the harm #Google is purposedly doing to #teens they intentionally addict, I would not touch their products with a stick.
Don't be their drug dealers.
Spreading their tools harms people.
@npub18s7…mtel

2026-02-05 17:52:16 UTC

> The documents provide smoking-gun evidence that Meta, Google, Snap, and TikTok all purposefully designed their social media products to addict children and teens with no regard for known harms to their wellbeing, and how that mass youth addiction was core to the companies’ business models. The documents contain internal discussions among company employees, presentations from internal meetings, expert testimony, and evidence of Big Tech coordination with tech-funded groups, including the National Parent Teachers Association (PTA) and Family Online Safety Institute (FOSI), in attempts to control the narrative in response to concerned parents.
> “These unsealed documents prove Big Tech has been gaslighting and lying to the public for years
https://techoversight.org/2026/01/25/top-report-mdl-jan-25/
#BigTech #GAFAM #Google #Meta #TikTok #Snap #teens #mentalhealth #privacy #surveillance #addiction #drugs

2026-02-03 12:13:05 UTC

- reply

Please invert the default: let people who want #AI enhancements enable them in the settings.
@npub1a3j…xsms

2025-11-19 07:49:12 UTC

- reply

With #snac2 you can host your own fediverse instance on the cheapest shared hosting.
https://encrypted.tesio.it/2024/12/18/how-to-run-your-own-social-network.html
@npub12pw…zvxn

2025-06-05 06:24:25 UTC

Working at #Google in 2000s by @npub18zk…45sm
Pretty interesting read that shows clearly how, as @npub12pw…zvxn often remind us, "all struggle is #class struggle".
https://wordsmith.social/elilla/deep-in-mordor-where-the-shadows-lie-dystopian-stories-of-my-time-as-a-googler
#privacy #capitalism #work #humanrights
/cc @npub1652…09vs

2025-05-10 10:32:19 UTC

- reply

I think this would be a much appreciated patch... except I guess it is javəscript based, isn't it?
@npub1mhv…8s5z @npub1x2y…8qaj @npub13qr…egvc @npub1mhv…8s5z

2025-05-10 08:28:02 UTC

- reply

in fact it doesn't cause any issue: #snac2 is likely the lightest and best performing #fediverse server out there.
@npub1h7c…smmd @npub13qr…egvc @npub1mhv…8s5z

2025-05-10 08:24:21 UTC

- reply

because xs_list and xs_dict are just byte array under the hood.
@npub1h7c…smmd @npub13qr…egvc @npub1mhv…8s5z

2025-05-10 07:55:02 UTC

- reply

To be honest, xs (and xs_json too) is infact what I liked most of #snac2 codebase: a non-conventional approach optimized for the specific use case.
Having said that, it looks trivial to modify _xs_json_load_lexer, xs_number_new, xs_number_get, xs_number_str to store/retrieve the double as a null terminated char[sizeof(double)+1].
I just wonder if it's worth the performance gain, tbh.
@npub13qr…egvc @npub1mhv…8s5z @npub1x2y…8qaj

2025-02-28 08:28:56 UTC

- reply

"tanto rumore per nulla"?
Nelle [FAQ](https://www.mozilla.org/en-US/firefox/faq/ ) #Mozilla ha anche [rimosso la promessa di **non** vendere i dati degli utenti](https://github.com/mozilla/bedrock/commit/d459addab846d8144b61939b7f4310eb80c5470e#diff-a24e74e4595fa85440a2f4e7e5dcfe68aba6e1e593aef05a2d35581a91423847L60-L65 ).
Non stai parlando di utenti paranoici, ma di informatici competenti.

2025-01-07 12:33:19 UTC

- reply

Actually that would enable #snac usage in a specific educational use case: schools.
I think kids and teens might be able to learn how to use social-networks in a defederated #fediverse instance, being able to do their typical errors in a protected environment.
However I know no school so brave to experiment on this, because of all kind of issues they might have to face, from sexting to bulling.
Yet, I think that it might help kids grow more aware of the risks and implication (and maybe even more #privacy aware).
On the technical side: you could have a defederated Snac if you make it available over a VPN or local network only. Also, I suppose that messing crypto keys would isolate the instance's inbox. All hacks, though.
@npub1640…sjc4

2024-12-17 23:36:26 UTC

Finally out a little #tutorial about running your own #fediverse instance on cheap shared hostings with @npub1ysc…ffqw's #snac.
https://www.tesio.it/2024/12/18/how-to-run-your-own-social-network.html
#permacomputing #retrocomputing #hacking #cgi