this post was submitted on 13 Jul 2024
131 points (75.5% liked)

Open Source

36318 readers
346 users here now

All about open source! Feel free to ask questions, and share news, and interesting stuff!

Useful Links

Rules

Related Communities

Community icon from opensource.org, but we are not affiliated with them.

founded 5 years ago
MODERATORS
 

Interesting history and analysis of SMTP's history. How can we prevent fedi and other open protocols from suffering the same fates?

(page 2) 41 comments
sorted by: hot top controversial new old
[–] [email protected] 12 points 9 months ago* (last edited 9 months ago)

(This is as much an answer to some of the comments already raised, as to the article - which like most such personal pieces has pros and cons.)

As part of a previous job I used to host email for a small business - this was about 15 years ago. I ended up spending several hours to a day a week working on it; apologising to users, tracing and diagnosing missing sent email and the endless, ENDLESS arms war against incoming spam (phishing was much less of a problem then). The trust from the company in our email operation was very poor and you'd regularly hear someone apologising to a customer because we hadn't contacted them, or answered their email. The truth is much was going astray and staff were relying more on the phone than email because they knew it worked. You might guess from this that I'm terrible at running an email system but I don't think I am. I started moving email back in the late 80s when Fidonet was the thing, so I have some miles travelled. Tools have improved a bit since then, but so have those used by the bad guys.

I still consider one of the best things I did for that company was move our company email onto Gmail Business (which was free for us as a charity) Every single one of those problems went away immediately and suddenly I had a lot more time to do more important stuff. I would never self-host email again despite running several personal servers.

Plenty of people say they self-host just fine, and great for you if that's so. But the truth is you won't always know if your outbound mail silently gets dropped and you have a far higher chance of it arriving if it comes from a reputable source. There are a huge number of variables outside of your control. (ISP, your country, your region, your software, even the latency of your MX or DKIM responses factor into your reputation)

You take the decision on whether any perceieved risks of privacy through using a third party outweighs the deliverability and filtering issues of self hosting, but please don't say it's simple or reliable for everyone. If it's simple for you, you're either incredibly lucky or just not appreciating the problem.

[–] [email protected] 96 points 9 months ago (16 children)

You can't successfully use a home email server.

Mostly true (server can be home but using the ISP network directly probably won't work)

You can't successfully use an email server on a (cloud) VPS.

Bullshit

You can't successfully use an email server on a bare metal machine in your own datacenter.

Bullshit

As such, it is my distinct displeasure to declare the death of SMTP. The protocol is no longer usable. And as we can see, this devolution occurred organically.

Bullshit

[–] [email protected] 38 points 9 months ago

You can’t successfully use an email server on a bare metal machine in your own Datacenter

Calling complete BS on that. I work in a medium size company and we do just that. Don’t know what he’s thinking.

[–] [email protected] 1 points 9 months ago* (last edited 9 months ago) (7 children)

Sure, you can run one, good luck getting even a halfway decent delivery rate to mailboxes at any major mail provider. Even if they never receive a spam message from your server, your server is an "unknown" which counts against you. And if one person in your small company of 10 or 100 or even 1000 people gets their e-mail hacked and sends spam? Prepare for the rest of them to get punished for it. Running an SMTP server is a nightmare which is why, over time, more and more of the economy has just shifted their SMTP servers to organizations who professionally run SMTP servers instead of having their own.

[–] [email protected] 7 points 9 months ago

I've never had any issues getting mail delivered to major providers

[–] [email protected] 20 points 9 months ago

I work as a Sysadmin for a web host who sells VPS's. I've helped many people setup domains on their server to cover SPF, DKIM and DMARC passes on a daily basis. Most use these for personal or business level mail delivery without issue.

Are there hurdles to overcome? Sure. But it's not exactly hard as long as you have a IP that's isnt a poor reputation (which as an ISP we help delist and improve). But it's not impossible.

Its more "convenient" to use a third party mail provider just as Office365 since you pass on all that setup and responsibility onto their framework, but it's not hard to setup a decent level of mail service yourself.

[–] [email protected] 1 points 9 months ago

You're spot on, and even smaller ISPs routinely get blocked by larger hosters (anyone who doubts this, please look around for the many stories along the lines of "gmail silently drops my email")

Residential IP blocks are scored much higher and given a negative trust from the start - not surprising since that's where much of the world's spam comes from through compromised computers, routers etc.

[–] [email protected] 33 points 9 months ago

Set up dkim/SPF properly, make sure the ip you plan to use is clean before you start, sign up for MXtoolbox blacklist alerts and if you get on a blacklist (doesn't happen often if you do a bare minimum of proactive security), you request removal. It's really not hard.

[–] [email protected] 26 points 9 months ago (1 children)

I am running my own mailsever for over 10 years without any blacklisting problems...

[–] [email protected] 12 points 9 months ago* (last edited 9 months ago) (3 children)

Right, but try doing that with a 10 day old server created in 2024. That's the hurdle people are referring to.

[–] [email protected] 14 points 9 months ago

This works, too. It's actually common that your dmarc-entry needs some time to be accepted everywhere. Wait a few days more and your mails don't hit the spam folder on google and outlook.

load more comments (2 replies)
[–] [email protected] 9 points 9 months ago

have mine running for nearly 20 years now and never had any major issues with delivery 🤷

load more comments (1 replies)
load more comments (14 replies)
[–] [email protected] 15 points 9 months ago

i can't read anything that's presented with that shitty cover image without a hint of irony

[–] [email protected] 26 points 9 months ago

I should have expected the rug-pull at the end when I read:

You may know me as a Bitcoin educator and engineer

However, I was still surprised!

[–] [email protected] 19 points 9 months ago (2 children)

Defederating bad actors/spammers should in theory be good enough? Domains aren't free and I don't think it's worth it for them to buy a new domain to just be able to spam for a short time again.

[–] [email protected] 32 points 9 months ago* (last edited 9 months ago) (1 children)

Domains aren’t free and I don’t think it’s worth it for them to buy a new domain to just be able to spam for a short time again.

Literally what e-mail spammers do.

Agreed defederating can help solve obviously malicious instances, it doesn't solve spammers abusing good instances. E-mail and AP are very similar at a protocol structure level.

[–] [email protected] 15 points 9 months ago (3 children)

Is it though? Don't email spammers just spoof the domain or send without a domain? I'm not entirely sure if that's different from how the fediverse works. I'm not too knowledgeable about this topic.

[–] [email protected] 5 points 9 months ago

You need to set up dkim to prevent spoofing. Each message sent has a digital signature that matches one on a DNS record for your domain. You can also set an SPF record, which will tell the recipient what up addresses are authorized to send mail on behalf of your domain.

The recipent must have policies in place that reject mail which fails dkim/spf

[–] [email protected] 12 points 9 months ago

Don’t email spammers just spoof the domain or send without a domain?

Very much so. Out of the spam that I do see in my inbox, the sender domains are usually spoofed, while the reply-to addresses are usually gmail.com, hotmail.com or outlook.com.

[–] [email protected] 4 points 9 months ago* (last edited 9 months ago) (1 children)

Don’t email spammers just spoof the domain or send without a domain?

They do both, depending on the spammer and the type of spam they send. In e-mail, you have an e-mail server, you can use it to send mail to users on other e-mail servers. Each e-mail server can choose to accept or reject email from other e-mail servers based on whatever reason they want. AP/Lemmy/Mastodon is basically identical to this. I'm not sure how exactly bluesky is setup but I get the impression it's similar. In Nostr, servers aren't federated (each relay is seperate, if you want to send/recieve content to another user on a different relays you just talk to that relay directly instead of having "your relay" act as an intermediary), but the structure is still pretty similar.

Nostr does have this hashcash type system (requiring proof-of-work to weed out spam), but I haven't come across any relays that actually enforce it, it will be interesting to see if that changes in time. I also saw a GitHub issue about adding something similar to AP but I think they chose not to implement it.

[–] [email protected] 8 points 9 months ago (2 children)

Replying to your edit:

it doesn’t solve spammers abusing good instances

This is an instance moderation problem. If you're letting spammers in, you need to use a better application process or something similar to that. A big problem with email spam is that most email services allow anyone to sign up for free without any checks.

Ultimately defederating bad actors and defederating "good" actors who fail to moderate their own users is necessary.

[–] [email protected] 2 points 9 months ago (1 children)

Ultimately defederating bad actors and defederating "good" actors who fail to moderate their own users is necessary.

Agreed, and this is what makes the Fediverse so good. It would be annoying to lose your instance, true, but you just move to another or roll your own. Additionally, let's say they start spamming Mastodon from mastodon.social; their messages would go to the Global channel, but if I only ever read Local or Subscriptions, I'll never see their spam.

The Fediverse and ActivityPub will continue to evolve, but unlike SMTP, they were created after the internet became adversarial. This author isn't the first to try to fearmonger over the future of AP, and they won't be the last.

[–] [email protected] 3 points 9 months ago* (last edited 9 months ago) (1 children)

It would be annoying to lose your instance, true, but you just move to another or roll your own.

This is a problem nostr solved, and I believe bluesky solves as well though idk as much about the protocol. On nostr, your identity and your instance are different things. Relay goes down? There's no meaningful impact to you. You're typically connected to several, each of which store your content. You identity isn't username@somerelay dot com, it's just username.

As a user, I had this happen to me early in mastodon and it was very frustrating to lose all my follows, followers, tweets, settings, etc. I realize there's now ways to manually backup etc but properly moving an account requires a cooperative instance which can't happen if it's de-federated or just drops offline randomly like mine did.

The Fediverse and ActivityPub will continue to evolve, but unlike SMTP, they were created after the internet became adversarial. This author isn’t the first to try to fearmonger over the future of AP, and they won’t be the last.

This isn't fearmongering, it's him reviewing the ways SMTP tried to solve the spam problem and became centralized as a result. These questions of how we tackle spam and moderation are valid, important questions. And Fediverse, at a structural level, is basically the same as SMTP. We have users at instances (e-mail hosts), they can send messages/tweets/links (emails) to users on other instances. Each instance is free to accept/reject messages from other instances based on their own criteria. That's the whole thing. That's exactly how SMTP works.

[–] [email protected] 4 points 9 months ago

It is fearmongering, albeit unintended, but I don't think it completely applies to the Fediverse as it stands. We should always remain vigilant and never complacent, and I'm sure the devs and moderators are keeping spam control in their minds. This isn't the 1980s, and we're not trying to retrofit a protocol that came before spam was ever a thing.

[–] [email protected] 7 points 9 months ago* (last edited 9 months ago) (1 children)

This is an instance moderation problem. If you’re letting spammers in, you need to use a better application process or something similar to that. A big problem with email spam is that most email services allow anyone to sign up for free without any checks.

Which is one reason, this author is arguing, that e-mail has become so centralized. Doing that kind of manual moderation and curation is expensive, the bigger instances out-compete the smaller ones who don't have as much resources to dedicate to it. As more and more instances get "de-federated" for not having as good of anti-spam measures as the bigger instances, more users will sign up at big instances to avoid defederation risk. Just like how many people use gmail simply because their email delivery rate is so good. If I send from g-mail, there's very few servers which will reject my message or throw it in the spam folder. I'd love to run my own mail server, but even as a dedicated sysadmin it's impossible to get decent delivery rates.

The more anti-spam checks we have, yes we weed out spam, but we also make it accessible to less users as well.

AP has been blessed so far with not having to fight too much spam. Look at very popular, very centralized, very resourced platforms like Facebook, spam is still a problem on their platform despite massive resources put towards fighting it.

[–] [email protected] 6 points 9 months ago (1 children)

Hmm I feel like some pooling of effort with spam detection built into the software (lemmy for instance) could help spread the effort of spam fighting to other, smaller instances and not just centralised to the big ones.

But it's difficult to say what will happen I guess. We need to just keep being vigilant when it comes to stopping spam while keeping in mind our shared goal of a decentralised social Internet.

[–] [email protected] 3 points 9 months ago
[–] [email protected] 5 points 9 months ago (1 children)

But most people don't pay for software, especially if there are "free" and legal alternatives.

[–] [email protected] 8 points 9 months ago (1 children)

I'm not sure what you mean with that or how it relates to what I said, could you elaborate?

[–] [email protected] 2 points 9 months ago (1 children)

As I understood you said that we should make email paid to stop spammers.

[–] [email protected] 7 points 9 months ago (1 children)

Nono, I'm saying it costs to spam because spammers have to keep buying new domains as their previous domains get blocked or defederated.

[–] [email protected] 2 points 9 months ago (2 children)

Why don't they use existing domains? I don't think 100% of them require a phone number. And didn't know it's possible to defederate an email provider.

[–] [email protected] 2 points 9 months ago

And didn’t know it’s possible to defederate an email provider.

It absolutely is, your mail provider "de-federates" aka blocks mail from plenty of other e-mail providers.

[–] [email protected] 5 points 9 months ago (1 children)

No, my point is that if spammers were to spam on the fediverse, they'd need to buy new domains constantly as their previous domains are defederated, I'm not talking about email.

[–] [email protected] 0 points 9 months ago (1 children)

So you're offering a system that requires the instance/provider admins to manually federate with others instead of the federation being enabled by default?

[–] [email protected] 4 points 9 months ago (1 children)

You're misunderstanding me again. Please try reading what I said again.

I'm not suggesting allowlist federation, though that is another tactic that could be used. I'm just saying that a spammer on the fediverse would be quickly defederated and would have to buy a new domain to keep spamming, which would probably be too expensive to justify.

[–] [email protected] 4 points 9 months ago (1 children)

Yes, mister/miss, I read your comments again and I admit I did completely misunderstand you. I'm sorry for that and I'm extremely grateful you chose to explain your point in a friendly manner instead of starting a drama or acting in a toxic manner.

[–] [email protected] 4 points 9 months ago (1 children)

You're welcome m8, have a nice day

[–] [email protected] 4 points 9 months ago

You too, mister/miss

[–] [email protected] 50 points 9 months ago* (last edited 9 months ago) (1 children)

I don't buy this. I'm still using SMTP on my own domain and it’s working fine, a bit of spam but not unmanageable, real messages get read. Main challenge is digesting so many potentially-interesting list messages, indicating email's continued dominance for professional topics. Seems this author has another agenda.
Having said that, it's a pity the world never agreed a protocol for micro-payment for emails (and for many other services), which would resolve the spam problem, and not be a burden for honest users.

[–] [email protected] 5 points 9 months ago

I also host my own mailserver and I agree that it mostly works fine. However, there are some email providers that cause trouble:

Google seems to randomly sort some of my mails into the recipients spam folder, while others are delivered fine to the respective inbox. It kinda sucks that you can never be sure whether the recipient actually received your mail or whether they just don't reply. My IP and domain are not blacklisted on any spam list; SPF, DKIM and DMARC are set up correctly as well.

Even worse is the Telekom (German ISP), who use an explicit whitelist of IP addresses (only IPv4 of course) and require you to display your contact information publicly on a website reachable via the same domain your mailserver uses. Once you've set this up you need to message them to be put on their whitelist. If you're not on their whitelist, they simply reject your mails, they are not even delivered to the spam folder (maybe it's not worse than Google, because you at least get a notice from your mailserver that your mail couldn't be delivered). In the end I decided that I don't care enough to comply with their regulations and just don't send any mails to Telekom customers.

Aside Google and Telekom, I've really never had any issues though.

load more comments
view more: ‹ prev next ›