self hosting is the only way to guarantee logs arent shared.

however it also means that your server address only pulls requests for things YOU look at, not the whole pool of people using the server. imo this is a bad trade off as it would allow the services to track my browsing easily anyway.

personally I would recommend looking at libredirect with a large # of frontend instances selcted & a vpn/tor. these websites normally work without javascript, so assuming you have a privacy focused browser your IP is the only tracking metric they really have. spreading your requests out over different servers whilst using a proxy would make this really difficult to correlate as it would require that mutliple instance operators were malicious & collaborating.

common aegis W

you can use something like smspool.net to verify the account with a one time use #. then google will allow you to link an authenticator app to the account. removing the # from the account afterwards does not remove the TOTP code either.

you can try looking it up on plexus to see if others have had success using that app

if you want to degoogle the only way to run some apps without unprivileged code (GMS/microG) is to use grapheneos & its sandbox compatibility layer. this is unfortunately only available on the pixels.

I also worry you won't be able to use some of the customization options that the nothing phone 2 offers with its hardware when using a custom ROM. if thats something u care about.

I have the pixel 7a & have had essentialy no issues with anything for what that's worth.

ahhh I see what you mean.

your thoughts on spacing out your connections & isolating is smart. unfortunately if you connect from the same device & browser any government agency or dedicated company with a big enough dataset (google, meta, etc.) would still be able to link you regardless of you IP by browser fingerprint alone. this does make YouTube more specifically being linked to your exact browser fingerprint porblamatic in a high stakes situation. As it, as you said is linked to your identity.

for lower level tracking changing IP regularly is effective. however, instead of switching to your local IP it would be more privacy conscious to just switch to a different VPN server.

unfortunately if you are genuinely worried about government level surveillance or the likes u enter into territory where VPNs often no longer cut it (or at least can't truly be trusted too) as they are centralized & can be forced to make exceptions for law enforcement. traffic analysis is also easier, which makes time correlation deanonimization a more realistic risk when talking about government agencies specifically.

the tor + vpn debate is one that lots of people argue & is excedingly complicated. tor is generally more than enough, unless you are wanted by INTERPOL haha. if you are genuinely worried about suppressive government or world powers targeting you look further into tor, & do not connect directly to your ISP at all as that data is essentially up for grabs to local authorities (depending on locale).

for you specifically I would consider doing your more sensitive tasks in the tor browser without the VPN & then having your normal browser always on the VPN so they would be more difficult to correlate. anything torrent related is low enough stakes that I would imagine just about any proxy would suffice. hope this was helpful 🙏.

tldr; no, if you trust your vpn more than your ISP always use it, as any hit to fingerprinting is menial.

it really can't hurt much to always be using it. any fingerprinting metric it would give is outweighed by the hiding of your IP behind the proxy. this is the #1 unique identifier that is tied back to people/locations.

the other fingerprinting metrics also are still exposed anyway & could probably be linked back to "you" regardless of your IP changing if they wanted too.

if you are worried about fingerprinting look into some projects like mullvad, librewolf, or even tor. clearing cookies on quit &/or having a separate browser for permenant logins/tokens to live in is also a good mitigation technique.

important second frame for context!

& no it isnt. quite sure twitter broke link previews a long time ago alongside guest accounts.

this is extremely scary if true. are these algorithms obtainable by every day people? do they work only in heavily photographed areas or do they infer based on things like climate, foliage, etc? I would love some documentation on these tools if anyone has any.

outside of sandboxing/limiting android auto via grapheneos good luck.

if you just want audio & your car doesnt support Bluetooth you could use an adapter to take it from Bluetooth to 3.5 aux. if your car is so fancy it doesn't have aux input you can get a Bluetooth to FM radio adapter that will take over one the channels in your car as well.

scrambled exif is really practical & easy to use. it has been a player for a while now where as metadata remover seems newer & possibly less trustworthy/vetted because of it.

it also doesnt seem to add much new functionality from scrambled exif besides viewing data in app.

I personally use image toolbox which allows you to select what specific exif tags you would like to remove. however, I wouldnt put any more trust in its accuracy or comprehensiveness than any other adjacent app as they realistically all probably work very similarly.

if its imperitive you don't leak any exif data I would recommend checking that in an external & more trusted app anyway. something like aves or a similar gallery that will lay out all all metadata tied to an image.

unified push works as a stand in for gms on devices without it. it runs in the background & receive the wakeup pings for the apps (in this case simplex) so you only need one websocket open instead of a different background service for each app. hugely reduces battery use.