DuckDuckGo costruisce il suo indice web: meno Bing, più AI
DuckDuckGo sta costruendo un proprio indice del web. Lo hanno annunciato il fondatore Gabriel Weinberg e il CTO Caine Tighe in un episodio del loro podcast interno, spiegando le ragioni tecniche e strategiche dietro una scelta che arriva quasi vent’anni dopo la nascita del motore di ricerca.
La storia è curiosa: DDG aveva già iniziato a indicizzare il web nei suoi primissimi anni, quando Weinberg lavorava da solo. Poi aveva abbandonato l’idea, preferendo appoggiarsi a indici di terze parti, Bing su tutti, e concentrarsi su quello che aggiungeva valore sopra: risposte istantanee, knowledge graph, ricerca locale. Una scelta pragmatica per una piccola realtà che non poteva permettersi l’infrastruttura di un grande motore di ricerca.
Oggi le cose sono cambiate, e la spinta principale viene dall’intelligenza artificiale. Search Assist, il sistema di risposte automatiche integrato nei risultati di ricerca, e Duck.ai, il chatbot dell’azienda, hanno bisogno di accedere a contenuti web freschi e controllati direttamente. Avere un indice proprio significa poter alimentare questi strumenti senza dipendere da dati di terzi, con un ciclo di feedback molto più stretto: i milioni di utenti che usano DDG ogni giorno diventano, in forma anonimizzata, un segnale continuo sulla qualità dei risultati.
Tighe descrive la pipeline tecnica in modo abbastanza dettagliato: c’è un crawler che rispetta le preferenze dei siti (chi vuole essere indicizzato e chi no), un sistema di rendering che esegue anche il JavaScript delle pagine per estrarne il contenuto reale, e un motore di ricerca semantica basato su embeddings (rappresentazioni matematiche del significato dei testi) che usa un database chiamato Vespa. L’indice è già attivo su una parte del traffico e, secondo quanto dichiarato, sta crescendo giorno per giorno.
Vale la pena notare che questa è comunicazione aziendale e le motivazioni dichiarate sono credibili, ma il fatto che l’AI sia il motore principale della decisione merita una lettura attenta: DDG si è sempre posizionata come alternativa rispettosa della privacy, e costruire un indice proprio per alimentare prodotti AI è una direzione che porta con sé domande legittime su come quei dati vengono usati internamente. Per ora non ci sono ragioni per dubitare della coerenza con la loro politica di non tracciamento, ma è un’evoluzione da seguire.
Chi usa un motore di ricerca che non vuole essere tracciato da Google o Bing, e magari abbina già una VPN come Proton VPN per proteggere il traffico di rete, ha interesse a capire come si muovono i player alternativi. DDG che riduce la dipendenza da Microsoft è, sulla carta, una buona notizia. Quanto questa indipendenza si traduca in qualcosa di concreto per gli utenti lo vedremo nei prossimi mesi e anni.
FONTE insideduckduckgo.substack.com
Luca Sironi
in reply to Elena Rossini 🌈 • • •ah, ecco, mi ero perso l'account di Henna Virkkunen
@europeanspodcast @haubles @aral @Gargron @tferrer @rstockm @samvie @EUCommission
Luca Sironi
in reply to Luca Sironi • • •@luca
I think I'm gonna ask all the commissioners with a #bluesky account, to *at least* enable #bridgyfed
@quillmatiq
@_elena @europeanspodcast @haubles @aral @Gargron @tferrer @rstockm @samvie @EUCommission
Oblomov reshared this.
Michel Patrice
in reply to Elena Rossini 🌈 • • •Also, X/Twitter give you the illusion of being able to take part in a discussion. Unless you have a the blue check mark, no one sees what you write and no one sees the comments you adress an elected official using Twitter.
Imagine a political assembly where only those wearing a paid badge can use the microphone.
Moving public deliberation to Twitter undermines the very foundation of democracy.
Em reshared this.
Michel Patrice
in reply to Elena Rossini 🌈 • • •Nicolas Fressengeas
in reply to Elena Rossini 🌈 • • •@europeanspodcast @haubles @aral @Gargron @tferrer @rstockm @samvie @EUCommission
> I look forward to hearing your thoughts and suggestions
Well, great article, Elena ! I would love to disseminate a French version.
Did you link with @leavex ? They carry roughly the same message though not ruling out Bluesky.
Thank you for your thoughts about Bluesky, by the way! I am struggling to understand eurosky. Any thoughts on it?
#SocialMedia #DigitalSovereignty #BigTech #Fediverse #RSS
Ricardo Antonio Piana likes this.
Leave X - Protect Democracy
in reply to Nicolas Fressengeas • • •@fresseng the #LeaveX campaign doesn't rule out #Bluesky, but we favor the Fediverse and particularly #Mastodon. See our message:
leavex.eu/posts/concrete-europ…
We discussed this with @_elena during #FOSDEM. Specifically, we believe that the Fediverse is the most resilient solution. However, we don't want to exclude the possibility of building *bridges* with the millions of people who have chosen Bluesky. (cc @bjoernsta)
@europeanspodcast @haubles @aral @Gargron @tferrer @rstockm @samvie
Concrete European alternatives to X: Mastodon and Open Portability
Leave X - Protect DemocracyMichel Patrice
in reply to Elena Rossini 🌈 • • •I don't get the rss part.
I tried with your adress in the Thunderbird rss reader thing, it doesn't seem to work. I added /rss to your adress, still doesn't work.
Does it work with any fediverse adress, or does the account need to have somehow enabled somekind of rss switch?
luca
in reply to Elena Rossini 🌈 • • •Michel Patrice
Unknown parent • • •When I will get back home tonight, I will try with some other accounts.
This rss feature would be quite cool.
ImaCrea
in reply to Elena Rossini 🌈 • • •nicolas ⁂
in reply to Elena Rossini 🌈 • • •al
in reply to Elena Rossini 🌈 • • •👏👏
Ruud
Unknown parent • • •nicolas ⁂
Unknown parent • • •Saw it there so it’s still pretty early I’m guessing: github.com/mastodon/mastodon/p… 😅
P.S. Oh no, too bad you’re missing on the conference! ☹️ Are there maybe any other events you’ll be going to? 😊 Pretty sure I’ll attend, yes!
P.P.S. It’s actually public just as of today!! Have a look here for a first glimpse at it: code.vinyl-cache.org/vinyl-cac… 😁 (By the way, they’re on Mastodon too now! And thank you again for sharing the original call for entries! I’ll post an announcement soon as well)
Add email subscriptions by Gargron · Pull Request #38163 · mastodon/mastodon
GitHubAquaClaire
in reply to Elena Rossini 🌈 • • •This is such an important issue! All universities, institutions, citizens & responsible communicators should be using a "communications platform that is accessible to all citizens, without the need for an account; an independent network not subject to [monetisation &] censorship due to opaque algorithms or political bias."
Thanks for this clear explanation!
#Fediverse #privacy #PublicGood
Nuvalon
in reply to Elena Rossini 🌈 • • •you make some really good arguments, so i might share this with someone that works for the city.
Michel Patrice
in reply to Ruud • • •@ruud
It works! You have to add .rss instead of /rss.
It is set up on my Thunderbird. I will see, Elena, if your next posts do appear in my rss feed.
Leonieke
in reply to Elena Rossini 🌈 • • •Bart Knubben
in reply to Elena Rossini 🌈 • • •🙏 Thanks for making the case!
I believe it is also important to learn from and ‘fame’ the 🇳🇱🇩🇪🇫🇷🇪🇺 government organisations and public institutions who are already active on Mastodon/Fediverse 🧐🙌
mastodon.nl/@bartknubben/11614…
Bart Knubben
2026-02-27 09:02:03
Elena Rossini 🌈
in reply to Elena Rossini 🌈 • • •Yesterday I wrote: "Public institutions are funded by taxpayers' money: their communications serve public interests and should be open to all - without the need of creating social media accounts on proprietary, for-profit closed systems. This is especially salient for European governmental communications: isn’t it absurd that they would require social media accounts owned and run by U.S. based companies?"
Credit goes to @samvie for suggesting I add the bit "funded by taxpayers' money" 🙏
reshared this
Roberto Resoli, Maho 🦝🍻 e Oblomov reshared this.
John Faithfull 🌍🇪🇺🏴🧡✊🏻✊🏿
in reply to Elena Rossini 🌈 • • •Talia Hussain
in reply to Elena Rossini 🌈 • • •government is not “funded by taxpayers money”, too bad you’re spreading this pernicious myth with an otherwise good argument
Better to say “public money” which would be correct and accurate.
Icanbob
in reply to Elena Rossini 🌈 • • •Robert
in reply to Elena Rossini 🌈 • • •social.overheid.nl - Mastodon Overheid
Mastodon op social.overheid.nlmacfranc
Unknown parent • • •Michel Patrice
Unknown parent • • •It also works with hashtags. I just tried it.
Look for a hashtag in the search box. Click on the hashtag in the result list. Copy the link in navigation bar and add .rss.
I am now just waiting to see if I will get this hashtag's publications from the whole fediverse (which would be cool) or from Mastodon only.
Oliwier Jaszczyszyn
Unknown parent • • •@_elena: mind if I make up a Polish translation and publish it on kontrabanda.net?
@macfranc
ImaCrea
Unknown parent • • •Souverain 🛡️
in reply to Elena Rossini 🌈 • • •Souverain 🛡️
Unknown parent • • •Souverain 🛡️
Unknown parent • • •Souverain 🛡️
Unknown parent • • •Souveraineté numérique des partis et mouvements politiques français : analyse comparative des dépendances aux GAFAM et grands intermédiaires
SouverainSouverain 🛡️
Unknown parent • • •Souverain 🛡️
Unknown parent • • •Souverain 🛡️
Unknown parent • • •On my end, on the Fediverse, I found the following accounts:
@renaissance
@renaissance_78@bird.makeup
@re_thiais
@renaissance@twitter-bridge.cryobyte.net
@renaissance_fde
@renaissance_62
@renaissancehdf
@enmarcherbt@bird.makeup
@ren_arras@bird.makeup
As for bluesky, I agree that 36 followers is truly ridiculous, but I would be dishonest if I said it didn't exist.
And as for "Joli bullshit de l'opposition" don't worry, you're not old school because : "il faut appeler un chat, un chat" 🤣
Souverain 🛡️
Unknown parent • • •Souverain 🛡️ (@souverain@social.souverain.ovh)
Souverain 🛡️