Hacker News new | ask | show | jobs
by timdorr 3447 days ago
Since those are accomplished via a tracking pixel in the email and you need to serve that image from a public server, then yes, you'll need a cloud service to run that. As stated before, you can choose to run that server yourself if you don't trust Nylas with that kind of data.
4 comments

Or you could implement RFC 3798, an update of RFC 2298. It's not like these things haven't been standardized for almost 2 decades.
Unfortunately, standardization != widespread use. Tracking pixels are still the most reliable way to know whether a recipient (or many) opened your message.
Unless your recipients use email clients that don't load remote http images except on demand.

I can think of one email client with ~15 million users that does that.

Which client blocks remote images by default? Is it a cli or gui?
Gemail, Outlook.com/hotmail, thunderbird and outlook all block images, unless whitelisted... Now, there may be other providers/clients that don't, but the above accounts for a significant number of users (if not most western mail users).
Gmail does something more interesting. It shows the image by default if the url is not unique, or the image is an attachment. It also uses googles servers to download and cache the image when receiving it.
"When you get a message that has a picture in it, you see the picture automatically. "

https://support.google.com/mail/answer/145919?co=GENIE.Platf...

Alpine, gui of sorts

K9, gui Android (think it blocks by default, or I turned off auto load when I installed ages ago).

Thunderbird, gui.
I think he meant Gmail
Every mail client I've used, including webmail, has blocked external images by default for at least the last decade.
I believe Exchange/O365/Outlook block by default when the senders are outside the organization.
That's not a good excuse for not implementing a standard. It's also a little like saying "private investigators are still the most reliable way to know whether a person is at home": it's true, but not really a better thing to implement.
Whether a standard has adoption or not is a perfectly good consideration when deciding whether to implement it.
Neither MS Outlook nor GMail use a tracking pixel.

Such pixel are only used by newsletter or shaddy spam senders.

Tracking pixels are ubiquitous, and used in virtually almost every situation where you want to know if a person opened your email, which equates to virtually all ecommerce in my experience.
As you said, in newsletter/spam business. But common mail apps don't do that - there is already a defacto standard for that. It's clear that Nylas business is around tracking and collecting data. It's like Win10, the user is the product and everything gets collected.

Good that Outlook and other mail apps blocks third party pictures by default.

There are other mail clients that use tracking pixels, like Polymail and Canary.
The parent mentioned probably the two biggest mail clients in the world, and you mentioned 2 that I haven't heard of despite 20 years of sending email and toying around with new software.. I'm not sure it's an effective rebuttal.
So this won't work with emails coming from Gmail (and probably other providers) since these services cache these images on their server as soon as they receive them (and not when recipients open them) precisely to defeat this kind of tracking.

And of course, a lot of email clients only open such images on demand anyway.

I don't believe this is correct. Gmail will only request an image to proxy once you've opened the email, according to MailChimp, so this would only prevent tracking multiple opens.

http://blog.mailchimp.com/how-gmails-image-caching-affects-o...

It's actually hard to track down exactly when these images are cached. I don't really trust MailChimp to be truthful about this but I also can't find any specific from Google themselves about whether that caching happens when their SMTP receives the email or when the user actually opens the message.

Since Google's goal with this was clearly to defeat tracking, I would strongly expect the former, but I can't back this up.

I haven't been able to find the tracking pixel server on GitHub.