LibRedirect – Redirects popular sites to alternative privacy-friendly frontends | Hacker News Reader

LibRedirect – Redirects popular sites to alternative privacy-friendly frontends

libredirect.github.io

424 pointsriffraff102 comments

Overall it works but the problem lies in instances that tend to die-off pretty fast. There were homebrew "hubs" solely providing redirects out of pure kindness to many big sites and services but now it seems it's hard to find one that works without being blocked/rate limited. Big sites and services fight back, which isn't really surprising.

Privacy Redirect was prob the first extension that introduced this idea. It did the job as well but up until bad-actors figured out they can redirect people to their dangerous sites.

Seems related, so I’ll share here. I wrote an “awesome” list of privacy-focused front-ends[1] for a variety of services. Haven’t been updated in a while, but I figured it’s still valid.

[1]: https://sr.ht/~jamesponddotco/awesome-privacy-front-ends/

Instagram doesn't actually work, right? All frontends are down, and it doesn't seem to work locally either.

One word - Imginn

That is correct, there's no alternative front-end that still works. Self-hostable open source ones, that is—you can still find random ones on search engines that aren't open source.

gallery-dl still works, but doesn't have a web frontend.

imginn works for now

when I see my partner doomscrolling, I think that is for the better.

I just found out about an Android app where you can set up custom redirects for any links, OS-wide: https://github.com/TrianguloY/URLCheck

It's a little finnicky to set up, but I'm enjoying it so far. It goes beyond alternative frontend redirects. You can strip URL params, check domains against a blacklist, and choose native apps to open links that match a pattern.

I was very happy when I found about that app, It's very useful. It goes beyond just redirects, it's able to remove tracking elements from links, unshorten links, remember which app to use to open specific domains and more. You almost need an app like this on Android because of its shitty share menu.

Nice, I was looking around for something like this and found this Android app instead: https://codeberg.org/hermeticvm/linkahest

I started using this last week and it's simple and useful. My main challenge was that I use YT/Reddit/X apps but I hate how the recipients I shared links to posts or videos with often couldn't access it without having to disable ad blockers or having to login first.

Redirector[1] makes it easy to set up your own redirects. I prefer that.

[1] https://addons.mozilla.org/en-US/firefox/addon/redirector/

A web extension is an unnecessary security risk. A userscript will do it just fine.

edit: one of my previous attempt: https://news.ycombinator.com/item?id=35229211

I actually have made it extensible, with closely coupled source of rules and domains; but then I lost it Edge forgot all my userscripts :(

User scripts have super wide permissions. For example a user script scoped to YouTube.com can make payments from any cards you have saved in Google pay.

And most user scripts are so long a typical user won't be able to spot a couple of malicious lines amongst 10k lines of minified webpacked libraries.

> And most user scripts are so long a typical user won't be able to spot a couple of malicious lines amongst 10k lines of minified webpacked libraries.

Exactly!

That's why you should use 3 lines for it instead, that are

   - inspectable
   - not updateable by the Chinese/Russians
   - written by you anyway

You also have to weight the benefits versus the "risk".

For example, if you use FreeTube with SponsorBlock to improve your privacy and block ads, in fact you are sending to Cloudflare 100% of your YouTube watch history, and to SponsorBlock ("sponsor.ajay.io").

With Piped instances it's even worse, essentially escaping Google's tracking just to give our data to random strangers.

If you are worried, just run a second Chrome session with NordVPN and uBlock Origin in a loose jurisdiction and browse YouTube unlogged.

It's easy, simple, and you have the benefits of an audited platform and that reasonably legally confirm they don't store logs unless the court forced them: "we never log their activity unless ordered by a court never log their activity unless ordered by a court", but for that, the court has to find you as a user, which can be very complicated in practice.

So much better than random strangers.

> If you are worried, just run a second Chrome session with NordVPN

I feel like I’m on YouTube already.

It’s not like they are free of criticism either.

https://en.wikipedia.org/wiki/NordVPN#Criticism

I'm happy to give my watch history to some unknown in exchange for never ever seeing an ad.

>If you are worried, just run a second Chrome session with NordVPN and uBlock Origin in a loose jurisdiction and browse YouTube unlogged.

If you actually did this you would know that it works for all of a week or two before YouTube stops letting you watch videos until you login.

I found that hopping to different VPN servers is a mildly inconvenient workaround for that.

SponsorBlock doesn't send video IDs to the server.

https://github.com/ajayyy/SponsorBlockServer/issues/25

(*anymore, as of late 2020 from a quick look. The parent comment may not have been wrong about that, just outdated info)

Terrible advice. Not only youtube will precisely fingerprint you, nordvpn/tesonet/oxylab will also get data on you.

Way better than the recommended "privacy" instances.

NordVPN only sees that you connect to YouTube, they do not see the pages or videos that you are looking at, and from the perspective of YouTube, they only see requests from a very popular VPN where are millions of users.

If you use the "privacy" instances, these "privacy" websites and Cloudflare knows precisely which videos you are watching.

Recommended by whom? I'm just saying your advice is terrible in general and takes no regard to how easy and powerful fingerprinting is nowadays, in google's perspective the only difference to using that VPN if you're "just" running chrome is that it also knows when you use a VPN, in other words, just giving one more data point. Also the average user is likely to install some nordvpn app if following your advice, which is a security nightmare, remember that company sells residential proxies.

Also IIRC for youtube, alternative frontends don't tend to rely on someone else's endpoints.

> worse, essentially escaping Google's tracking just to give our data to random strangers

I'd much rather send random tidbits of information, that are nearly useless in isolation, to strangers than to the central tracking corporation

In the end, there is no way to reveal what information you're interested in when retrieving data, short of retrieving a ton of data and doing the filtering client-side, which is also an option with these third parties if you so desire

The extension links to 50+ services, your script - to 1. Do you now suggest that every single user should figure out how to do it properly and replicate the extension in a script for no better alternative (you could instead spend part of that time reading the extension code and using your private copy)

I don't think that not having all the services is a problem. On the contrary, I think it is an advantage for userscripts, that those only have the redirects a user explicitly adds.

Tho I probably should've demonstrated first that it is possible, before advocating for it. The script I linked indeed only works for one website. Multiple websites with multiple rules, each with a list of instances (that often go offline for a time, so it is worth keeping them around, and make switching easy) indeed complicates it a bit.

So what exactly is the advantage of having to code all the rules yourself for every service you want to use??

> complicates it a bit

a bit of an understatement

> So what exactly is the advantage of having to code all the rules yourself for every service you want to use??

"having to code all the rules" is not that hard, in most cases you can just pass the whole URL, and the instance accepts it.

Advantages: you don't get unwanted redirects from services, and you don't get unwanted redirects to instances. (Even tho the information about the instances will likely be concentrated at libredirect github issues. Chances are that some random person on the internet who has paranoid activities as a hobby will look into the instances, so you don't have to.)

- - -

I don't use many redirects. Nowadays I use exactly 0. But if I needed a redirect for example to xcancel, I would use my user-script as I had done it in the past before I lost it. I definitely wouldn't install a browser extension for it.

> in most cases a slice(,) will do it since the relevant id is at a fixed position in the URL.

In all cases that also involves actually finding the URLs, then there are non-most cases where a slice wouldn't do it.

> Nowadays I use exactly 0

Exactly. If you ignore actual uses everything becomes trivial

just disable auto update and have the same bad usability as user script.

I personally prefer to use redirector to do it. It has served me quite well so far.

https://einaregilsson.com/redirector/

Totally unrealistic. Instead either lock down extension permissions, use different browser profile or better yet use QubesOS for spinning up disposable browser VMs

can a userscript run before the page loads...? afaik it's not possible, so the browser gets to make double requests.

the privacy stuff is fine(if not a bit suspect since we're still relying on the goodwill of the instance hosts to not be sketchy) but for me the biggest benefit for these third party front-ends is that my crappy laptop isn't constantly being pushed to the limits of it's capabilities just so I can read some gosh-darn text.

And reddit is not even close to the worst offender in that regard. Seriously, when did displaying words on a screen become so resource intensive???

have you tried old reddit?

The no-JS / reduced-JS aspect of some of these frontends is particularly interesting, since it implies that the JS wasn't ever necessary --- except perhaps its only purpose was to be privacy-invasive and user-hostile.

I reckon a very large chunk of the JS on many mainstream websites is explicitly for data collection and advertising.

Nobody is setting up "privacy-friendly" frontends to track browsing data that they couldn't otherwise get without access to Google's/Twitter's/etc. logs? Because I think they are.

Yeah, the possibility of any of them being a honeypot I'd say is real.

How could you ever prove that nobody is doing that? You can believe anything that way

One can't prove god doesn't exist either, but as someone who made some privacy-friendly front-ends, I tend to expect honest intentions. If you find one that suddenly asks for your login data or sets tracking cookie, sure, be wary, just as with any other site that asks for data they don't need (see: literally every cookie wall, because if they had good intentions, it would fall under one of the five other reasons to use personal data and they wouldn't need to fall back to asking for consent)

Nothing. An acquaintance of mine develops a third-party frontend explicitly marketed as a privacy-friendly alternative and actively looks at lots of user data (which includes the full name) without disclosing. I honestly believe that it's only done for improving the service (and it helps tremendously) but I can't get through with arguing that this should be transparent.

You could notice by closely reading the source code.

Don't use it. Stop shitting on everything you disagree on. Besides, privacy is not black and white. No one is implying such a ridiculous claim. Just because you grew up in a disgusting for-profit driven web, doesn't mean that everyone is trying to get you. Believe it or not, there are people who actually value privacy and actively voluntarily support decentralized and non-invasive parts of the web without hoping for any incentive. Besides, majority of private frontends are extremely fast and loads in an instant, which saves a lot of time

They are all effectively proxies so you do have to trust them to some extent, but unless these frontends are run by a large company, I think they couldn't care less - and likely don't even have the resources to accumulate and analyse all the data that passes through them.

This is just going to normalize adware / phishing. These front ends can show ads or ask for users personal information.

Redirecting people from trusted sources to these other sites is very risky and opens up opportunities for malicous people to exploit this. That's not even considering this extension is compromised or purchased and these dangerous permissions that it has are used against you.

Yeah, but these front-ends mostly aren't built for authenticated usage. The tracking done by the actual source sites is quantifiable and immense. Proxying is a good option for removing a lot of the tracking capabilities, especially via an instance you control.

>an instance you control

But most people are not going to do that. They are going to be redirected to a site with no guarantee of what is there. The domain could expire and someone else could register it, a hacker could replace one of the front ends with a phishing page, etc.

I reckon one of these front ends serving ads will have trouble maintaining a user base large enough to justify the effort of serving ads

Do any of these YouTube extensions retrieve videos in a way which is unassociated with my IP? I’d really rather not get my google account banned, or my searches rate limited. These aren’t happening now, but I believe they will in the future to the point where I actively avoid using any tooling from my home connection, and vps’ seem to be blocked by YouTube already.

If you have a dynamic IP at home, run it in your homelab and access it through Tailscale everywhere. I highly doubt YouTube will block the whole IP block for home users.

That doesn’t solve the issue of my google search traffic and fingerprint from coming from the same source as yt-dlp.

VPNs are not blocked by YouTube.

Neither is viewing YouTube using Tor Browser.

Thank you.

At a glance, it does not seem like they fixed the main issue with what they forked: Not being able to set up arbitrary targets.

I run my own instances for a few of the services they redirect to, and need to be able to point to these.

The YouTube alternatives always lag and are bad, unfortunately. I don`t know why.

The best way to watch YouTube videos is actually to download them with yt-dlp then watch with mpv later.

I want the opposite, an extension that will redirect all crappy frontends to the canonical sources (which work better and I am logged-into, I can comment, etc).

Don’t almost all of them show a link to the source anyway?

So... press the 'clone' button on the repository and swap the mapping from twitter.com -> nitter.net to nitter.net -> twitter.com?

It would be nice to have a containerized host of all these services to have them easily on hand as needed. One more task for Claude to handle...

Any good YouTube options (including self host)? I’ve tried a few and they always seem to be down more than up.

Did you have a look at peertube ? https://joinpeertube.org/en_US

I did, seemed to fall in the same category of sometimes working, sometimes not. I'v been trying various alternatives on/off for the past 5 years or so but unfortunately nothing really ever sticks.

Thanks for your feedback

https://grayjay.app/ perhaps? It's a locally running application. Don't know how privacy friendly it exactly is, but they claim they collect very little information.

Farside extension, 847 stars: https://github.com/benbusby/farside

Using venrable farside.link

https://sr.ht/~benbusby/farside/

https://farside.link/

Why use your offering?

Maybe for the fact it as 4 times as many stars on GitHub if that's what you care about?

This comment could've been phrased better, but Farside does have an important feature that LibRedirect lacks, which is automatic instance selection based on reachability. Instances routinely fail and new ones are added, so automating that aspect instead of requiring manual instance selection by the user is a powerful feature.

Anyway, thanks for mentioning it!

Using Farside means the initial redirect goes through Farside, so they are capable of knowing what videos you're watching, what tweets you're looking at, etc. You have to trust them not to monitor this. Using a client-side extension means only the instance you use knows this.

It's a Go project that seems trivial to self-host. By your logic we shouldn't trust any of the instances of the alternative services either since anyone could be monitoring their use as well.

Removing telemetry from daily tools feels like taking back a little control every time.

X.com works bet with lightbrd.com instead of xcancel with captchas.

lightbrd also needs cloudflare captcha

Try nitter.tiekoetter.com.

That also uses a captcha, not Cloudflare, but Anubis

I have never seen an xcancel captcha..

Neither do I - just the usual "verifying your request" screen: https://i.ibb.co/MyWRVtFj/xc.jpg

Which is a PoW CAPTCHA, but a CAPTCHA nonetheless.

However, if your JS is disabled (or if you're running LibreJS), you do get redirected to a CAPTCHA which only works sometimes.

Just nitter.net is better than both of them.

How long before browsers disable these kinds of in-user-favor workarounds?

Like Apple removing the "Disable JavaScript" menu option from Safari and moving it into Developer Tools, which can be detected by websites before you can disable JS >:(

I think the real question is: should we keep using browsers that are developed by ad companies? And the answer is no, we should just use Mozilla Firefox.

We should all use Tor browser alongside Firefox.

Download today people https://www.torproject.org/download/

Firefox should be replaced with an actual private clearnet browser like mullvad browser that is using most of Tor browsers protection without the mozilla tracking and ads.

Mozilla is an ad company now: https://www.mozilla.org/en-US/advertising/ https://blog.mozilla.org/en/mozilla/mozilla-anonym-raising-t...

This is not at all comparable with the big companies.

> Like Apple removing the "Disable JavaScript"

That's so fucking good. I love the cope from Apple users

I love this extension

Proxigram? I doubt I could run that on Android.

...care to elaborate why you can't visit a website on Android and how this is relevant to anyone else?

It is on the list of "LibRedirect", and it seems to be a self-hosted front-end to Instagram, not something one could just simply download from F-Droid and use.

Oh you mean that it's a website and not downloadable software, right

Yeah, I thought I found a FOSS, easy-to-use frontend to Instagram that could replace the Instagram app. :/

thank you so much brother, the info is very helpful.

Looking under the hood, some of these things seem like they might be moving your data from one place that might not have your best interests to another place that doesn't seem to have a revenue mechanism?

Take for example nitter - it says its using an unofficial twitter API. I'm assuming this means its using one of these third party services that provide an API to something that doesn't necessarily have an API or has limited access thereto.

If privacy is the purpose, this seems to be missing the point.

> Take for example nitter - it says its using an unofficial twitter API. I'm assuming this means its using one of these third party services that provide an API

You misread that. It actually says:

  Uses Twitter's unofficial API (no developer account required)

In other words, it's an internal Twitter API that's not meant to be used for applications like this.

time to get rid of the freeloaders with Anubis? https://anubis.techaro.lol/

"privacy friendly". Now there's a modern euphemism.

What is implied?

Reply on news.ycombinator.com