## 2023-03-17

### arrl.net and the ARRL -- or: How Not to Treat Your Members

(I apologise for the inconsistent size of the font in the main text of this post. I wasted twenty minutes of my life trying to force blogger.com to eliminate most of its auto-generated convoluted HTML so that I could then manually rationalise the look of the text, but I was forced to admit defeat. I refrain, with some difficulty, from ranting about the blogger.com interface for creating blog entries.)

## The arrl.net Service

The February, 1999 issue of the ARRL's journal QST, included the following item (the above link may require membership in ARRL for access) in the Happenings column:

[I note that the ARRL attempts a limitation --

-- that is grossly over-reaching and, in particular, makes no allowance for fair use -- a legal concept that, while a little blurry around the edges, would certainly apply to, as here, the duplication for noncommercial purposes of a small part of one column on one page of a magazine of well over 100 pages.]

Like many active hams, I saw the value of being able to use an e-mail address of the form <callsign>@arrl.net for day-to-day e-mail, and I signed up for the service.

For more than twenty years, the address N7DR@arrl.net has been my usual e-mail address, used for communication with friends, family and many other categories of correspondent.

[In passing, I should note that a couple of people have recently pointed out that the above announcement made no mention of sending e-mail from the arrl.net domain. However, the distinction is rather disingenuous. In 1999, there was no expectation, or even general notion, that a valid e-mail address could not be used as a matter of course for both sending and receiving e-mail. Indeed, even now, it is hard to see the practical purpose of an address that can be used solely for receiving personal e-mail on a day-to-day basis. Had there been any expectation in 1999 that an address would be so limited, it's hard to envisage many members signing up. Or, indeed, that the ARRL would not have included a specific warning not to use the address for sending e-mails. And, in practice, the arrl.net addresses worked perfectly fine from the time of the service's introduction for both sending and receiving communication. And it's also difficult to understand how it could have been otherwise: the technical method that was eventually introduced to implement the limiting of e-mails in this manner was not even introduced by the IETF until the publication of RFC 4408 in April, 2006 -- and even then the RFC was explicitly stated as being experimental (i.e., explicitly not an IETF standard).]

So, for more than twenty years, arrl.net addresses worked as one would expect: e-mails sent to your arrl.net address would be forwarded to a server from which they could be downloaded using whatever protocol(s) that server supported (typically POP3 and/or IMAP); and, if you configured your e-mail program (technically, a Mail User Agent, or MUA) so that the From: header contained your arrl.net address, you had every expectation that the associated e-mail would be delivered correctly by whatever SMTP server one used to relay outbound e-mails.

And then came 2023, and an unannounced policy change by the ARRL (at least, I have not been able to find any warning, explanation, discussion or announcement of the change -- or even an after-the-fact description of it, to say nothing of any kind of motivation for it).

## The Symptom(s)

So what were the symptoms of the problem that arose because of this change?

The first indication of a problem I saw was that an e-mail for a gmail.com user was bounced by that domain's e-mail server. Now, for gmail to bounce a legitimate e-mail is hardly unknown [as far as I can tell; despite the fact that they almost certainly scan the contents of every e-mail they handle, they don't actually use that information for determining whether a given e-mail is "spam" -- technically, what we generally refer to as "spam" is "Unsolicited Commercial E-mail", or UCE)]. Historically, gmail would occasionally decide that all e-mails from the server through which I route outbound e-mail is UCE, even though a glance at any of my personal e-mails would have been sufficient to determine that, while they are indeed e-mails (!), they are generally not "unsolicited commercial" e-mails. Anyway, every now and then, gmail would bounce my e-mails for a day or two before suddenly allowing them again. This would happen perhaps every year or so, usually for a couple of days, before gmail would revert to accepting e-mails from the server. So it was annoying, and demonstrative of sloppy coding, but not really a major issue.

But now the symptom was different. The bounce message from gmail.com said:

  550-5.7.26 The MAIL FROM domain [arrl.net] has an SPF record with a hard fail
550-5.7.26 policy (-all) but it fails to pass SPF checks with the ip:
550-5.7.26 [64.62.234.38]. To best protect our users from spam and phishing,
550-5.7.26 the message has been blocked. Please visit
550 5.7.26 information. g32-20020a9d12a3000000b0068cd21180c7si2618456otg.140 - gsmtp

As usual with bounce messages from gmail, the included link was useless. But the text of the message is sufficient to describe the problem in detail: the SPF record for arrl.net suddenly was indicating that a legitimate e-mail from the arrl.net domain could come only from particular pre-defined servers. First, what exactly is an SPF record?

RFC 7208, published in April, 2014, is the IETF standards-track document defining the Sender Policy Framework (SPF). The SPF is in tended to allow the owner of a domain to place restrictions on the servers from which e-mail claiming to be from that domain are allowed to be received. In other words, a company can use SPF to inform the recipient server whether an e-mail purporting to come from its domain as actually being received from a server that is permitted to send e-mails from that domain. Great for a company... but hardly the milieu in which the ARRL's e-mail system operates, as there are hundreds (thousands?) of members legitimately sending their @arrl.net e-mails from a multitude of equally-legitimate SMTP servers, none of which are known to, or under the control of, the ARRL

## Sender Policy Framework (SPF)

SPF is defined in RFC 7208; a good less-technical explanation of the parts that matter to us may be found here.

The basic idea is that the recipient e-mail server may (there is no requirement that it do this, and many e-mail servers are configured not to do so) issue a Domain Name System (DNS) query against the domain of the e-mail's From: address. The response from DNS may include an SPF record (again, there is no requirement that such a record be included in a domain's DNS configuration).

So, suppose that an e-mail server receives an e-mail from N7DR@arrl.net. It may query DNS and receive, as part of the response the following entry:

arrl.net        text = "v=spf1 include:spf.protection.outlook.com include:pobox.com -all"

This tells the e-mail server that the owner of the arrl.net domain (i.e., the ARRL) has included a record complying with SPF version 1 (as of this writing, that is the only extant version).

Skipping some details, the two "include" fields tell the server to query those domains as well; typically, that query informs the e-mail server that arrl.net emails received from machines in those domains are to be accepted. Unless one happens to be using outlook.com or pobox.com as one's outbound e-mail server, these "include" fields may be ignored for our purposes.

The important part is that last field:  -all.

The above-referenced document describes that field well. It says, in part,:

"-all " is the part of the record that indicates what is recommended to do if the sending IP address does not match any of the ones in the record. This is determined by whomever publishes the SPF portion to the DNS record, such as the owner of the domain.

So, in our case, this is what the ARRL has decided should be done with @arrl.net e-mails that do not come from outlook.com or pobox.com.

The document continues by describing what that field means:

1. Types of rejection levels:
• -all (reject or fail them - don't deliver the email if anything does not match)
• ~all (soft-fail them - accept them, but mark it as 'suspicious')
• +all (pass regardless of match - accept anything from the domain)
• ?all (neutral - accept it, nothing can be said about the validity if there isn't an IP match)
Most records will have a "~all" listed in the SPF record because the domain owner leaves room for the possibility of a new server getting created and might forget to update the SPF record with the new IP address of that server. This also allows for regular machines to send email without causing too much of an interruption.

Very large domains such as gmail.com have "?all" in their records to leave it up to the recipient to determine what to do with the email when received.

So, since the ARRL has no idea about which machines are originating the e-mails from all their members who use @arrl.net addresses, the one thing they should NOT be doing is to set that value to "-all". As the ARRL really can't say anything about the legitimacy of any particular @arrl.net e-mail "?all" would seem to be the correct value; but "+all" would seem to be a reasonable alternative; less reasonable would be "~all", since a recipient system might well regard anything marked as "suspicious" as to be treated differently from normal e-mail, and there is really no basis for suspicion -- certainly no basis for the ARRL to suggest a priori that it is a suspicious e-mail.

But, as we can see, the ARRL is in fact claiming that @arrl.net e-mails from other than outlook.com and pobox.com are to be summarily rejected.

No wonder, then, that suddenly e-mail servers are (surprise, surprise) rejecting legitimate personal @arrl.net e-mails.

Following the first appearance of a problem sending e-mails to @gmail.com accounts, I sent test e-mails to a number of other domains, and found that results were all over the show.

Some e-mails were delivered as normal. Presumably the destination servers at these addresses did not perform SPF checking.

Some e-mails were delivered, but with subject lines amended so as to indicate that the e-mail was spam, even though the e-mail could not possibly be regarded as being of an unsolicited commercial nature. [Why it is legal for a server to amend the contents of an e-mail subject line is a mystery to me, but apparently it is.] These servers seemed to perform an SPF check, but then ignore the ARRL's explicit instructions to drop the e-mail and instead treated the "-all" in the SPF record as if it were "~all". One can see the argument for this behaviour, even though it is non-conformant.

And some e-mails were simply silently dropped. Which has the sole merit of being in accordance with the ARRL's misguided instructions.

So the end result of the experiment was that, basically, an @arrl.net e-mail may or may not be delivered, may or may not be bounced, and may or may not be treated as spam. In other words, the address becomes, at best, unreliable -- and, in practice, more or less incompatible with ordinary day-to-day use.

So, what does one do in this situation? It turns out that what one does NOT do is to contact the ARRL to try to get them to fix this mess (of their own making).

## Interaction with the ARRL

Well, the niggling concern that there was no reason a priori to associate competence at dealing with circulation matters with that required for handling e-mail issues was quickly proven to be well founded. I'm sure that the person at circulation@arrl.net meant well, but unfortunately the best of intentions are no substitute for competence. A response was quickly forthcoming. Unfortunately, that response consisted of the following:

Thank you for reaching out to us.  I just sent  a test message and it did not
bounce back.  Please confirm receipt of that message in your xxx@xxxx.xxx account.

In other words, to respond to an issue regarding the sending of e-mails, they tested the reception of an e-mail. I explained the issue again, and this time a week went by without response. By this time I had worked my through the details of the problem as described above, so I sent a new e-mail, suggesting that the problem be put before someone who understood SPF records. In response, I was told merely:

I am following up with IT

with no further details, nor any explanation as to why I had heard nothing in the meantime.

Shortly afterwards, I received an e-mail from ARRL support at the rather peculiar and hardly-confidence-inducing non-ARRL e-mail address: support@22454831.hubspot-inbox.com, to confirm that a support request had been filed and it was being "reviewed". Like the e-mail from circulation@arrl.org, it gave no details as to the contents of the request for support.

So at this point, I was left dangling, with no real confidence that a competent support request has been filed. Nor with whom it had been filed. Nor how I could have filed a support request myself, thereby skipping the step that involved a person at ARRL who, it seemed, did not really understand the issue.

Your ticket has been closed. Thanks for contacting our support team.
We hope that your issue has been resolved to your satisfaction.

So: no information about the wording of the original support request contained (so I wasn't particularly confident that the problem had been reported correctly); no information about what the problem was determined to be; no information about how the problem was resolved; no information about any testing that had been performed to confirm that the issue had been corrected. In fact, not a thing about that brief communication seemed remotely commensurate with the professional competence that one would reasonably expect after reporting a technical issue.

Obviously, the thing to do was to test their "resolution" of the issue. So I sent an e-mail to a gmail account and... surprise, surprise, no change at all. The e-mail bounced, with exactly the same error message as before. Not to my surprise, looking at the SPF entry for arrl.net confirmed that no change had been made.

The e-mail from the support people contained a weird non-sentence that said simply "Request details" on a line by itself. I'm not sure how one is supposed to interpret those two words, but I took them to mean that I could request details of what had been done to resolve the ticket. So I sent an e-mail that included, in the relevant part, this text:

Yes, I request details. You say that the ticket is closed, but provide no
explanation of what you have done, or why you think that something has changed.

I just sent a test e-mail here, and the behaviour is precisely as I reported
before, and the problem remains exactly as it was.

As far as I can tell here, absolutely no change has taken place in the course
of the past week.

I don't intend to be rude, but I have received nothing from the ARRL that
gives me any confidence that anyone there either understands the problem or
how to fix it, despite my providing what I believe was a clear description of
the problem, along with its cause and what needed to be done to eliminate the
problem.

In response to that, I received exactly the same e-mail that they'd sent before saying when the original request for support had been filed from circulation@arrl.org:

Your request has been received and is being reviewed by our support team.

Four days later (you can guess where this is going), I received an e-mail precisely the same as the one I'd received before, saying:

Your ticket has been closed. Thanks for contacting our support team.
We hope that your issue has been resolved to your satisfaction.

And, of course, testing the situation revealed no change whatsoever. So I sent an e-mail to the support address, with a copy to circulation@arrl.org:

Why do I get the impression that there is no human there at all?

1. I requested details as to what action you had taken. You did not provide any.

2. You say: We hope that your issue has been resolved to your satisfaction.
But I already informed you that there was no change in status at all.

I just performed another test, and the problem remains. It is evident that --
if you changed anything at all -- you did not go to the trouble of testing
whatever you changed 🙁

Do I sound frustrated? I am. How do I escalate this to a real live human being
who can actually understand and fix the problem???
I never heard anything further. And have heard nothing further to this day. 

## The Work-Around

Given the utter lack of meaningful response from the ARRL, the last, desperate remedy is to purchase an account at one of the companies that are permitted by the Sender Policy record to forward e-mail from an arrl.net address. I purchased a basic account a pobox.com (\$20 per year), reconfigured my SMTP server to forward the e-mails from my arrl.net address to pobox.com, and, hey presto! suddenly I could send e-mails to all the addresses that had become problematic when I was forwarding through my normal service (I continue to use the original service for all my other (i.e., non-arrl.net) e-mail addresses).

Of course, this will last only until such time as the ARRL unilaterally changes its policy again and removes the "include:pobox.com" field in the SPF record; which, as precedent has shown, is liable to happen without warning or explanation at any time.

How most members using the service were ever supposed to figure any of this out is a mystery. As are so many things about this whole mess; springing immediately to mind, for example, are the questions:

1. Why did the ARRL decide to change the SPF record so as to eliminate all e-mail servers except those run by two particular commercial entities?

2. Who at the ARRL actually made that decision?

3. Why were those particular entities chosen?

4. Why was the change made without warning?

5. Why was the change made without any notification or explanation in QST?

6. Why was no detailed step-by-step how-to guide issued to help members understand what the ARRL has done and how to work around the problems their action has caused?

7. Why is the designated contact person for e-mail issues at the ARRL the person responsible for unrelated issues related to QST circulation?

8. Who actually is responsible for support of this service at ARRL?

9. Why is there a way in which ARRL staff can open support tickets, but no method is provided that allows members to do so?

10. Why are members not permitted to see the contents of support tickets filed by staff in response to issues raised by members?

11. Does the technical support actually do anything to fix problems, and, if so, do they test the solution? (Obviously, the answer was "No" to at least the latter question in my case.)

12. Why do e-mails from the technical support staff come from an address that has nothing to do with the ARRL (so that many people might easily automatically filter such e-mails unknowingly to a bit bucket without ever seeing them)?

13. When the technical support staff close a ticket, why do they not describe what they have done and how they have tested the purported solution?

14. Why do the technical support staff simply ignore requests from members for details as to the remediation they have performed when they claim to have fixed an issue?

I note that throughout the whole saga, not even a hint of a vague explanation was provided to me as to why the ARRL precipitated this problem. Maybe there was a legitimate reason. In fact, I hope there was.

(I note that the only reason that springs immediately to my mind is gratuitously pernicious: that they have put themselves in a position to offer a separate and distinct service to members to process outbound arrl.net e-mails -- thereby working around a block that they have themselves emplaced -- and for which they could charge an additional fee.)

I'm afraid that. as a long-time member of the ARRL, I find the combination of incompetence and arrogance displayed by its reaction to a member's legitimate issues with a service that has functioned correctly for more than twenty years to be inexcusable, even if it transpires that there was in fact a good reason for the change in their Sender Policy that precipitated all the problems. From my conversations with other amateurs preceding these events, it is clear that the ARRL has enough problems with its image and perceived competence among rank-and-file members (and non-members), without the need to create more.