In many ways I see this similarly to the consolidation of browsers, but less consolidated. I think about the advantages and disadvantages of the prominence of Chrome (65%), Safari (20%), Firefox/Samsung/Edge/Opera/etc (15%). With Chrome we’ve seen Google move the browser and related standards forward through sheer marketshare. CSS/HTML/JS standards live and die by Chrome support and that’s both good and bad. They have made great and opinionated strides when it comes to SSL/TLS. For example, Google effectively killed Symantec’s certificate business because it was mismanaged. They also effectively got rid of EV certs and pushed secure-by-default web server design where HTTPS appeared normal, but warnings all over the place for non-encrypted connections. On the other hand, Google is fairly disliked in the privacy community and those communities prefer independent Firefox. For email, I can see similar issues, mostly around security. If Microsoft were to decide security mechanism X is not worth the effort they can effectively decide to not implement it. What will internet users do, block all Microsoft email services? Conversely they could come up with their own security mechanisms and effectively force the rest of the world to adopt it. I do think centralization of email providers provides little potential for negative impact aside from operational issues. For example, outages probably have a wider impact due to number of users, but I can’t realistically see a scenario where Microsoft/Google does something “bad” with their email platform that affects the rest of the ecosystem. Caesar Kabalan From: NANOG <nanog-bounces+ckabalan=wlgore.com@nanog.org> Date: Tuesday, September 8, 2020 at 4:47 AM To: Mike Hammett <nanog@ics-il.net>, NANOG <nanog@nanog.org> Subject: Re: Consolidation of Email Platforms Bad for Email? I'm sure Dave Crocker has thoughts about this, but it has come up elsewhere. There are both positives and negatives about having such a consolidation. The positive is that it a small club can establish ground rules for how they will handle various forms of attacks, including BGP hijacking, DKIM, SPF, and other forms of validation to identify fraudulent mail, etc. Also, if you have a whole lot of postfixes and sendmails running around, that's a whole lot of code to patch when things go wrong. A small number of MSPs can devote a lot of time and paid eyes on code. They can also very quickly spot new attack trends. On the other hand, that means that it becomes difficult to become a new entrant, because one doesn't easily get one's mail accepted. Lots of grey/blacklisting (forgive the use of the term). Also, when one of those systems fails, it takes down a vast number of customers. Furthermore, it represents a massive concentration of private information that can be monetized. Eliot On 08.09.20 00:27, Mike Hammett via NANOG wrote: I originally asked on mailops, but here is a much wider net and I suspect there's a lot of overlap in interest. I had read an article one time, somewhere about the ongoing consolidation of e-mail into a handful of providers was bad for the Internet as a whole. It was some time ago and thus, the details have escaped me, so I was looking to refresh my recollection. Have any of you read a similar article before? If so, can you link me to it? ----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com Midwest-IX http://www.midwest-ix.com [https://www.gore.com/sites/g/files/ypyipe116/files/2017-03/Gore_logo_0.png] This email may contain trade secrets or privileged, undisclosed or otherwise confidential information. If you have received this email in error, you are hereby notified that any review, copying or distribution of it is strictly prohibited. Please inform us immediately and destroy the original transmittal. Thank you for your cooperation.

On 9/8/2020 10:59 AM, Matt Harris via NANOG wrote:

Once you get into that small club, it's just as hard to get kicked out, and unfortunately that means that if abuse, UCE, etc is coming from those hosts, they've got an even higher chance of hitting your inbox. So while in theory it might work the way you're thinking, in practice it hasn't because once you are in that club, a lot of the financial motivation to prevent abuse of your service - that is, inbox deliverability for your client base - goes away.

+1 Likewise, we're at a point now where if a criminal phish or virus comes from the largest few email hosters, and you provide them emails with full headers - the accounts do NOT get shut down. They literally don't think this is their problem. And likewise, data storage sites (GoogleDrive, OneDrive, etc) from the largest providers often will host malware for weeks or months without being shut down - or the malware at least persists for many days after being reported. The same is often true for their redirectors. Wwhat is frustrating is that the long-standing industry standard of "you're responsible both for what you both send and host - even if the malware wasn't intended" - seems to be lost. Likewise, back in the spring months of 2018, google's "goo[.]gl" shortner went crazy for a few months, and was being MASSIVELY abused by spammers, and was being used as an "end run around" URI DNSBLs (SURBL, URIBL, ivmURI, DBL). I collected 15K examples of abused shortners that were "live", and sent those to Google. At the time I sent those, only about 500 of that 15K had been shut down. What was infuriating was that 80% of these 15K shortners were pointing to only 12 spammer's domains. These should have been trivial to prevent! The OTHER infuriating thing was that my INITIAL response from my contacts at Google was - (I paraphrase) "other spam filters should just follow the redirect, and block these spams based on the URI it redirects to" - WOW! I sent them a very stern email about that. (and for comparison, abused Bitly shortners were mostly getting shut down within 2 hours - so "everyone does it" was NOT a decent excuse!) Like I said - the long-standing industry standard of "you're response both for what you both send and host - even if the malware wasn't intended" - seems to be lost on some of these large providers. Thankfully, this had a happy ending. After some "tough love" - Google replied back and said (I paraphrase), "we were planning on shutting that down - or at least shutting down the ability to add new ones - and due to your feedback - we're going to push that up a few months" - and so soon afterwards, they finally did terminate those 15K shortners - and stopped allowing new ones. So this is to Google's credit - but the problem had persisted for months - and it seemed like a lot of cultural/industry standards in the Internet Security industry seemed lost on them. Sadly, while this situation had a good ending - similar problems with the largest providers persist. At the same time, they sure can be draconian in how they block smaller providers who had a rare and short-lived security incident. The hypocrisy is incredible. For example, Microsoft will sometimes *permanently* block a small email hoster for a short one or two hour compromised email account situation that caused spam to be sent from that small hosters - but that was quickly fixed - even if that hoster sends MUCH legit email. It almost FEELS like extortion - since many of the IT people running those small-ish servers sometimes get frustrating - and move their email to the cloud - and then guess who OFTEN gets their email hosting business? -- Rob McEwen, invaluement

This is being portrayed a little too "either/or", that if you get spam etc from $BIGEMAIL you, service provider, block them. What goes on is multi-layer spam blocking using various tools rather than host/server blocking except as a last resort. So we'll block/toss/etc a lot of the malmail from $BIGEMAIL w/o generally blocking their servers. If we get a huge attack we have thresholds at which point we might block them for two hours (whatever) hoping it stops on its own or $BIGMAIL stops it. But those are pretty high thresholds and obviously can cause problems for our customers in delayed email but so can our mail servers being pounded on. Those $BIGMAIL delivery servers have a lot more computrons than we do. Aside: What's astounding to me is how little any of this has changed, other than consolidation perhaps -- remember when AOL's servers pounding you with spam could bring you to your knees? I do -- in over 20 years. -- -Barry Shein Software Tool & Die | bzs@TheWorld.com | http://www.TheWorld.com Purveyors to the Trade | Voice: +1 617-STD-WRLD | 800-THE-WRLD The World: Since 1989 | A Public Information Utility | *oo*

On Sep 8, 2020, at 4:38 AM, Eliot Lear via NANOG <nanog@nanog.org> wrote:

I'm sure Dave Crocker has thoughts about this, but it has come up elsewhere. There are both positives and negatives about having such a consolidation. The positive is that it a small club can establish ground rules for how they will handle various forms of attacks, including BGP hijacking, DKIM, SPF, and other forms of validation to identify fraudulent mail, etc. Also, if you have a whole lot of postfixes and sendmails running around, that's a whole lot of code to patch when things go wrong. A small number of MSPs can devote a lot of time and paid eyes on code. They can also very quickly spot new attack trends.

All true…

On the other hand, that means that it becomes difficult to become a new entrant, because one doesn't easily get one's mail accepted. Lots of grey/blacklisting (forgive the use of the term). Also, when one of those systems fails, it takes down a vast number of customers. Furthermore, it represents a massive concentration of private information that can be monetized.

You’ve also left out: Economic incentives to make questionable use of mail content and user data. Economic incentives to make life difficult for new entrants. Economic incentives to avoid transparency or convenience in addressing user concerns about erroneously rejected email. Reduction in consumer choice (if there are a handful of providers and they all provide essentially the same (crappy) level of service, then what can the consumer do about it? #include <std_list_of_oligopoly_prolems.h> Owen

Eliot

On 08.09.20 00:27, Mike Hammett via NANOG wrote:

...

I originally asked on mailops, but here is a much wider net and I suspect there's a lot of overlap in interest.

I had read an article one time, somewhere about the ongoing consolidation of e-mail into a handful of providers was bad for the Internet as a whole. It was some time ago and thus, the details have escaped me, so I was looking to refresh my recollection.

Have any of you read a similar article before? If so, can you link me to it?

----- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com <http://www.ics-il.com/>

Midwest-IX http://www.midwest-ix.com <http://www.midwest-ix.com/>

<OpenPGP_0x87B66B46D9D27A33_and_old_rev.asc>