Telemetry is fine when it is included with enterprise software with licensing agreements handled by lawyers and corporate security. Telemetry is actually helpful and a good thing in that use case.

It is not in consumer products, period. Unless you are paying me for this information (in actual money, not discounts, not services) telemetry should be banned.

Iftelemetry were so valuable (“data is the new oil”) the would/should pay us for it.

Like all those “We value your opinion” customer feedback surveys. Yeah, you value it at $0.

I absolutely disagree with these data collecting practices, but if your data is e.g. worth $5, that may have been already deducted in the printer's sale price.

If you have two feature-equal printers, but one doing data-collecting and $5 cheaper than the other, which one do you think will be sold out first?

I wish that manufacturers be forced to also show default/required data requirements on their products similar to how they already display minimum/recommended hardware requirements. This would at least increase consumer awareness of the issue, at best maybe abolish it entirely...

Printer hardware is usually sold at a loss or near a loss in order to focus profits on ink/toner sales. That's how you get the ridiculous DRM on ink.

Force manufacturers to allow third party ink and to disclose data collection and you get a much different, better for the consumer, printer market.

I don't want to be paid for it. I just want it to not happen.

Only when fully disclosed and blockable.

> ... with licensing agreements handled by lawyers and corporate security.

So it's helpful when it's someone else's problem?

It is helpful when the company doing the telemetry really values you as a customer and you have teams of people responsible for understanding and negotiating the details.

It is not helpful when the absence of competition forces you to accept one devil or another and little power (or time) to understand how your information is being gathered and used.

It's helpful when the iSeries calls IBM to get someone out for part replacement. Both parties know the whole deal since there is an actual contract with specifics. Consumers don't have 'contract support' unless they have cash.

> Consumers don't have 'contract support' unless they have cash.

So what do you propose then? Every website have a paywall and block poor people?

Plenty of people are happy to share information on the Internet and even to pay small sums of money to do so.

If you aren't interested in sharing information without selling visitors data, your service isn't viable without charge, or nobody is willing to pay you, everybody is probably better off without what you're trying to offer.

This isn't a website, it's a printer. We're already paying customers. If there was a paywall, we passed through it when we went to the store and dropped $100 on their product.

I suppose the big problem is that everyone gets into the data broker business when they get big enough these days. It fundamentally changes the expectations of your relationship with the company.

It's a lot like the Vizio/Samsung/etc Smart TV privacy fiasco. Back when you bought a $699 21" Zenith tube television, their business model was transparently "we make and sell televisions." The up-front cost was sufficient that they weren't too concerned with a trickle lifetime recurring revenue. There's no real place in that business model to focus on a data gathering side hustle, and you as a consumer had no reason to think they'd be interested that you kept the knob on UHF all night.

Similarly, if HP's business model is legitimately selling printers and printer accessories, there's very little information they need but are not getting from their existing "what retailers order for restock" and "direct sales and ink-as-a-service" channels. Even the obnoxious personalized 'you print lots of photos, buy our photo paper' ad doesn't require remote data submission; you could calculate it on the fly locally and pop up a banner, just like with 'you've printed 29 pages, time for a new cartridge!" I could see system and document info for crash log purposes, but even that's a one-time permission request you can make on demand.

I guess what's amazing is how much the tail has come to wag the dog-- they'd rather creep out people and run the risk people finding out losing the $100-plus-years-of-expensive-consumable sale in order to get that sweet sweet consumer-profile data worth a few dozen cents per-user in quantity.

Honestly, I want to replace my arthritic LaserJet 5 with something offering duplexing and more than four real-world pages per minute, but new printers seem to be doing everything they can to be a distasteful purchase instead of an exciting one.

Corporations with professional legal counsel have more power than individual consumers.

Realistically you want it at the router level. Force all traffic through a transparent proxy like mitmproxy with some custom code to strip out the sends you don't like. Stop all those IoT devices.

Its on my plate to make a go at it, with some inspiration from pihole. But really it'd be about enabling myself to use some of this great data without sharing it with a third party.

For example, I'd wear my fitbit if it wasn't reporting in to their servers. But if I force my phone through a VPN, which routes through my transparent proxy, I could feed fitbit junk data while scraping the pieces I want to my own system.

We need apps that take control of these devices and their telemetry away from the third parties.

Router solutions tend to work only for static/controlled lans. Nearly 100% of my time connected to a network is not under my control (someone else's wifi or telco). Another problem with router and other solutions such as pi hole / hosts is that they apply rules in a generic manner without regard to context. eg. On my Mac I use Little Snitch to disallow any comms with apple, except when it's App Store. Appalled at the telemetry that vscode allows, I've gone back to sublime text 2.

On my Android, (sadly without root) I use NetGuard for similar purposes. I blanket disallow google for many apps. I allow carte blanch to my personal servers for apps that I use, but any telemetry of theirs is stonewalled.

In Firefox I use containers to separate FB/Twitter to their own hole, while I blacklist them in uMatrix for every other circumstance.

That said, things like doubleclick and crashlytics are fine to be black listed throughout a network.

My thoughts were more targeted toward a properly sandboxed os that gives users the chance to control on a port/ hostname level what is being connected to.

> Realistically you want it at the router level.

I do both.

I run a firewall on my phone mostly to prevent applications from communicating out without my express permission. I also don't turn on my phone's radios without connecting to a VPN that I run at home, so that all of my phone's traffic gets routed through the defenses I've set up for my home network.

On top of that, I avoid using the web on mobile devices to the greatest degree possible.

Philosophically, I completely agree with you. The problem is that this isn't inherently possible because the whole thing could be E2E encrypted.

I know there is a Linux (python?) client that will sync (at least some models of fitbit) to their cloud service. But I've no idea if there is one that will dump the data locally. It's entirely possible that the cloud client is merely passing along an opaque blob.

Why not just block everything the printer sends? When does a printer actually need internet?

That's exactly what I do on my Mikrotik router. I assign a static address to my printer and then disallow any outgoing internet traffic coming from its internal IP.

Yes, removing Internet access is always an option (although eventually we might have to physically disconnect wifi antennas or use a Faraday cage...).

I was responding to a comment that was talking about creating Free Software to communicate with the device, specifically the idea of proxying access to the corporate server and modifying the communication, rather than implementing the whole protocol from scratch.

I'd guess the Fitbit protocol is encrypted, from a desire to keep people from cheating their activity reports. If a company wants to spend the development time, there is basically nothing that can be done to prevent a device requiring Internet access on a dumb-pipe all-or-nothing basis.

Although this comment may seem a bit extreme: I looked at the rubbish that Fitbit installed on my laptop several years ago. I decided to throw my Fitbit out after scrubbing my hard drive of their software. So, although it's fair to assume their software's probably improved since then, I'm not buying.

Faraday cage for the win. My printer stays offline - period. If something needs printing, I have to sneakerware it. If I'm unable to do that, I rethink whether it really needs printing in the first place :D

Hosts file blocklists work well for known domains. https://github.com/StevenBlack/hosts

Each application shipping with its own DoH resolver - spearheaded by browser vendors in the name of security - will put a lid on that.

They are rightly saying that DoH will secure requests from tampering --- but when it's the owner who's doing the tampering, it becomes yet another anti-user security feature.

Personally, I'm less concerned about privacy of DNS queries than the loss of control and need to have another centralised third-party in the process.

It was a bad security feature anyway, easy to bypass

On Firefox it is configurable or even disablable trivially. Much ado about nothing.

That doesn't mean the other apps which use DOH will also be trivially configurable to disable it.

This is exactly correct. The purpose of DoH is two-fold; for Google, to allow themselves to be the endpoint DNS resolver so they can both bypass local ad blocking and collect statistical behavioral data (I am aware that their policy does not permit them to do this; it will certainly not be the first time Google violates their stated policies in the mission of serving advertisements), and for e.g. Cloudflare, to centralize and control additional pieces of previously-distributed Internet infrastructure (and thus permit centralized monitoring not if but when they are compelled or subverted by the intelligence community).

Mozilla I don't understand. The most likely explanation appears to be that they are still in a catch-up-to-Chrome mindset, which is a disservice to themselves and their community.

DNS queries should be encrypted. Centralization-by-default is not the answer and people should look more closely at the incentives in play by those pushing the DoH standard. I appreciate the efforts of e.g. OpenBSD to prevent this side-channel leakage of user data to private corporations: https://undeadly.org/cgi?action=article;sid=20190911113856

> Mozilla I don't understand.

Mozilla also send tons of users data to Google, and, probably, gets money/better contracts/other benefits.

Proof: https://twitter.com/jonathansampson/status/11658588961766604...

A good DoH feature is that it distrusts the local network's services in favor of ones on the local device. Especially with Comcast, Verizon and others doing bad stuff with traffic, not trusting the network providers looks like a good thing.

Ideally one could change to any range of DoH resolvers - right now there's 3 or so.

DoT and running your own recursive resolver is better still.

Or if you work at an ISP, set up an alternative DNS that users can opt out of, for blocking telemetry for everyone except those that really wants it. Invert the playfield so to speak.

I'm glad I'm not the only one concerned about this.

As far as I'm concerned, Google are only interested in DoH so far as prohibiting DNS level adblocking within their walled gardens.

Everyone thinks they're working toward tamper proof connections for the benefit of users, but it's really for the ad companies IMO.

And manufacturers. I tried to install system-wide cert on my Android to intercept and see exactly what system apps on my old Nokia phone were sending to Chinese servers but couldn't because Google thoughtfully "protects" its users. Tivoization at its worst.

jup, I am concerned as well.

cloudflare DOH fortunately uses it own domain for dns, so you can block it at firewall level.

Google could be evil and make resolver "google.com", so you would have to block whole google.

I was talking to a few people, of creating a list of all public doh servers, so we could all use it on our firewalls to block them.

That doesn't help if ad companies and other attackers set up their own DoH resolvers.

That's why I've installed a MITM proxy for all HTTPS connections over my LAN.

Last time I put a large list of blocked domains in my /etc/hosts file, it was causing non-trivial amount of delay (hundreds of ms) to every dns lookup. I guess hosts file is not designed to be scalable. I ended up running a local dns server (which able to use those blocked domains list without noticeable performance hit). These day I just use pi-hole running on a spare raspberry pi.

Automatic blocking doesn't work cos they are constantly coming up with new ways to bypass blocks.

Here's what I use. There's a free tool called Windows Ultimate Tweaker. It'll help with basic settings.

Next, Du Meter - shows network traffic right on taskbar. If I'm not actively using the internet and Du Meter shows 1MB/s, I get suspicious.

Finally BWMeter. I'll say it's little snitch for Windows. It'll alert you any time an application tries to access the internet. You can allow/forbid temporarily or permanently.

They are all light on resources. BWMeter's UI isn't great but it gets the job done.

Anyone know Mac equivalents?

Litte Snitch

https://www.obdev.at/products/littlesnitch/index.html

And Activity Monitor in the Dock (Icon set to Network Access)

iStat Menus will give you amazing insights on your system in the menu bar. It has a rich choice of views.

There's also Little Snitch for network monitoring.

For the rest, haven't researched much.

Little Snitch -- https://www.obdev.at/products/littlesnitch/index.html

Not running any of bad oems specific software and relying on in kernel drivers to talk to whole classes of Hardware seems to work.

Pi hole, and similar, and block all other DNS, would probably be a good start.

Tackling on edge firewall, looking what goes through, and blocking it there is second step (but since a lot of it is going to various cloud providers and cloud flare) this is often not an option

I'm reasonably sure this stuff doesn't fly in the EU anyway (hiding away information like this deep within some privacy document is not the clear consent the GDPR requires).

Tools are becoming available though. Projects like PiHole are making it easier to block many malicious trackers. There are even companies selling pre-built PiHole devices. Unless HP is hardcoding IP addresses, it's only a matter of adding the required domains to a tracker blocklist (if they're not already on there) and most of these problems go away nearly instantaneously.

I've noticed my PiHole helping a lot in regards to stuff like mobile apps (Google Analytics, Facebook Graph, etc.) and embedded devices like these are probably no exception.

I can tell you that my UK printer has the same shit!

I went through it with a magnifying glass to make sure it didn't select anything.

In addition, I set up the IP stuff manually on the printer to ensure there was no gateway... can't get out without a gateway.

I used to do the same exact thing.

At some point, though, I noticed that "something" [0] still managed to "get out".

After running some packet captures, it became clear what was going on. Although the device was using the network settings that I had manually configured, I had not specified a default gateway. The device decided it would use DHCP to discover the default gateway for the network and began automatically using it so it could get out to the Internet.

Since then, I've started specifying a default gateway for any devices that I don't want to get out. I give 'em an IP address that isn't in use on the network and, fortunately, I haven't ran into any other instances of crap like this happening.

[0]: I really wish I could remember what device this was but it's been a long time ago and I really have no idea, sorry.

These kinds of concerns are why I put all my "untrusted" devices on a separate VLANs, so I can reliably shut them out of the internet. Simple VLAN-enabled switches don't cost that much any more. Such a switch allows you to treat any port of the switch as a distinct network interface on your main router, where you can just disable forwarding for selected interfaces entirely. It also prevents your untrusted devices from seeing each other, i.e. your printer wont't be physically able to send ethernet frames to your VoIP phone, even if connected to the same switch. Here is some introduction to the concept:

http://corecom.com/external/livesecurity/vlans.htm

> Simple VLAN-enabled switches don't cost that much any more.

They're dirt cheap and have been for the better part of a decade. If you're looking to upgrade, cost isn't a reason not to.

I did exactly that:

• http://wiki.alpinelinux.org/wiki/Linux_Router_with_VPN_on_a_...

• https://wiki.alpinelinux.org/wiki/Linux_Router_with_VPN_on_a...

My printer is in a VLAN which has no route out over the Internet. (The second link there uses VLANs).

True, but assigning it a static IP and blocking egress traffic originating from it should accomplish the same thing.

I think static IPs and blocking egress traffic only keeps well-behaved devices from doing any harm.

It does not protect you from compromised, malicous (IoT) devices. Think about a network printer doing ARP spoofing and MiTM-attacking your VoIP phone or IP cam. E.g. googeling immediately turns up vulnerabilities like this [1] one. A properly configured VLAN setup can help to prevent or limit this threat.

[1] https://www.scmagazine.com/home/network-security/hp-officeje...

New printer next year: "Hmm, I can't contact my telemetry server with these internet settings… time to connect to any open Wi-Fi I can find nearby!"

VLANs, huh? I'll have to Google that sometime. ;)

Are you up for filing a GDPR complaint with the Data Protection Authority? :)

I’d like to see a big hardware company explaining its shareholders that they got a fine of 4% of the company’s worldwide revenue because they wanted data that was supposed to increase sales.

I think we need laws.

Take a look at GDPR

I vote with my money. Laws are lazy, trivial to insert exploits into and stacked with unintended consequences. Eduication is much more effective.

Passing laws is pragmatic. Individual boycotts are idealistic, they go totally unnoticed and the status quo continues.

The "status quo" argument of apathy is self-fulfilling. There are few things that are more effective than market forces, hence the 24x7x365 war for your opinion.

Note how often people say "insert law here". Compare that with how many times they actually propose the text of a law. Our most effective laws are exceptionally simple and short. It's not accidental that "modern" laws are intractably complex.

How about try it? Propose the law.

Some people, dare I say most people, care more about results than dogmatic adherence to libertarian free market ideology. That is why effective regulation is everywhere you look. Regulation is popular and effective, regardless of how badly it offends your sensibilities.

Or? Is it in theory possible that your preception of reality is off?

Government regulation of the net (aka speech) is a non-starter, so the people who think "insert rule" fixes something are forced to rebrand it to "net betterness". More than half of the general public is wise to these techniques.

What "pragmatic" law do you have in mind?

Government regulation of businesses that happen to have websites (aka 'regulation of the net') is already reality. Pretty far from a "non-starter"

Surely you have atleast a sketch of the law in mind?

Hey I am interested in application level firewalls. Can you please share how you do this.

I think GDPR helps in some cases, but I suspect as the laws are analyzed and tested in court, the old habits will come back.

All OS vendors benefit from this telemetry, so they all have it and support it. Microsoft collects lots of data, but don't be fooled, Apple also collects lots of telemetry.

I think what folks will start to realize is that RMS was right and only free software will be the only way to navigate this mess (since users are not denied access to the source code, which can be analyzed and the idiocy removed, like people do with ubuntu).