Hacker News new | past | comments | ask | show | jobs | submit login
That time Verisign typo-squatted all of .com and .net (rachelbythebay.com)
298 points by zdw on Nov 28, 2023 | hide | past | favorite | 119 comments



Verisign continues to run a monopoly on .COM/.NET with basically no checks or balances to its power. ICANN has effectively surrendered to it. A significant amount of ICANN's budget is coming from Verisign. ICANN refuses to exercise any oversight, it's likely an example of regulatory capture at this point.

Verisign is going to push every possible way to extract more money from .COM/.NET monopoly. They maximize price increases at every opportunity on their contracts (which has automatic renewal and no competition). For what? No added value to customers or the ecosystem as a whole.

SiteFinder was just one of many moves in a long series of bad for everyone but Verisign moves. Remember when .ORG had its price caps removed and then private equity tried to buy it? Guess where a lot of that backing came from...


Richard Barnes wanted ISOC to get an endowment by selling .ORG. having .ORG is great for the nonprofits but really isn't something ISOC can leverage to support IETF etc.

I believe the registry contracts are bid out every few years. However it's a very difficult business with fixed prices that depends on lots of operational efficiencies.


Having .ORG was the endowment. Selling off something they don't/shouldn't 'own' is bullshit. They were granted an endowment to manage .ORG to use the profits to support IETF.

They simply outsource the backend and collect a large profit. It's not a difficult business at all. The margins are insanely good in fact. They're doing at least 60% if not closer to 90% I suspect on the actual operations side if they competitively bid out their backend.

Registry contracts aren't bid out every few years for the most part, that's a giant problem. Why Verisign has a monopoly on com/net.


I don't really understand why people care specifically about .com when they're still one of the cheapest TLDs you can buy. Like if you wanted to pin this on Verisign specifically you're gonna have to explain why every other domain costs about the same or more. Like tech darlings .io=$71, .tv=$32, .me=$25. Even the worst domain ever .info is $23.


Because .COM has history before Verisign. Because businesses and people are locked in forever to renewing. It's the defacto TLD for the world. Why don't you care that a monopoly contract that extracts profit for shareholders alone with zero public good is in the best interest of the internet as a whole? Why does Verisign of all entities get to manage it?

There are cheaper TLDs, everytime they go up for bid, the costs to run them comes out and it's under a dollar per domain. And you're comparing .COM an original gTLD to country code TLDs, which are owned by countries. Countries that can make their own rules and charge their own prices.


See also ‘domain tasting’ (https://en.m.wikipedia.org/wiki/Domain_tasting), where registrars would temporarily register non-existent domains when someone attempted to load them, showing a ‘for sale’ page to skim extra money off registrations.


Thinking of you godaddy.


Did GoDaddy do this, too? I only remember Network Solutions.

Incidentally, I was paranoid about this kind of front-running for years before I knew it had actually happened, and would just use whois from the terminal when searching for domain availability.


godaddy 100% did it. i remember looking for a domain, taking a day to debate, they reporteded it registered and they 10x'd the domain price on me. When I reached out to ICANN about such a bs move they just came back with "GoDaddy isnt doing anything wrong." This was 10-15 years ago.

That was the last time i registered anything with GoDaddy


I can't say I'm surprised, given that GoDaddy is the worst of the worst, from the sleazy pr0n Super Bowl ads to their (mis)treatment of customers.


Does any money change hands for this kind of front-running? If so, I think there's a good opportunity to either cost them a ton of money or drain whatever budget they allocate to this practice by baiting out registrations.


Originally it did, and some companies still tried the buy-from-under-and-jack-up-the-price thing on what could be valuable domains. Back then I checked the availability of a four-character (letter-letter-number-number) domain and two days later found it taken by “someone” using the same registrar I check it was free on and there was a holding page there offering it for sale at a not-insignificant cost multiplier. Luckily I had other options and just took one of those (from a different registrar). I also checked the availability of other domains on the original registrar, and encouraged others to also. We probably didn't cause enough financial disruption for someone to notice, but I liked the petty revenge anyway!

Later the 5-day grace period was added by ICANN to deal with accidental registrations, a full refund would be given if the domain was released in that time. Supposedly to protect end users against mistakes like typos and other errors, though I'm not sure why that would need five full days. This made “domain tasting” an open season and a great many registrars would do it, even registering a few times to extend the five days. Some actually did it as an advantage for the end user: they were not going to get snipped by waiting a few days and the registrar didn't jack up the price. But many were a bit more nefarious.

They later added a small processing fee to the refunds in the grace period after the first few domains per account per period (or similar) which vastly reduced this happening, so it is now pretty much a historic problem.


If I remember correctly (probably from stories on HN a few years ago), GoDaddy had the ability to "taste" the domains—ie, pay for them, hold them for a few days, then get a full refund if you didn't end up buying through them. I don't recall whether this was something special for large registrars like GoDaddy.


ICANN added a small nonrefundable fee many years ago that made this no longer practical on a large scale.


I'm sure tasting still happens on dropped domains, though; registrars have data on search interest and can find dropped domains that are likely to be profitable even taking transaction fees into account.


It's unfair that registrars can abuse their position to pick up valuable dropped domains. But somebody was going to do it in the moment after they drop if not them. That feels like a different and less serious problem.


Years ago I wrote a script that would release and renew a DSL line (to get random IPs) while spamming several registrars with bogus domain availability lookups, and found that something like 10% of the available domains were registered within 2 hours of looking them up, and I spammed something like 200,000 bogus domains in one night.


There was also the excellent (and more capable) bw-whois, which was retired in 2019.


From another angle, presumably they had some excellent engineering on their site to be able to handle the enormous up tick in server load following the change. A note from Wikipedia[0]

>According to the web traffic measurement company Alexa, in the year prior to the change verisign.com was around the 2,500th most popular website. In the weeks following the change, the site came into the top 20 most popular sites, and reached the top 10 in the aftermath of the change and surrounding controversy.

[0] https://en.wikipedia.org/wiki/Site_Finder


> BIND (aka named, the occasional remote sudo implementation)

Oh, goodness.

dnsmasq got a feature to block site finder - the "bogus-nxdomain" option - which still exists and still mentions 64.94.110.11 in the documentation.


"BIND (aka named, the occasional remote sudo implementation)" made me chuckle! I wonder if Paul Vixie still holds the record for the largest number of CERT advisories due to a single author?


It’s a fair assessment of BIND4 and BIND8, but BIND9 has been a lot more solid. (The delegation-only feature was BIND9 only; I can’t remember if it has been removed yet.)


> They could do this because they controlled those from the registry side of things, and it was trivial to slam something that would make it resolve.

Was this legal?

If so what stopped them there, they could just hijack any domain?


ICANN sent Verisign a strongly worded letter. Verisign stopped, but then sued ICANN and they settled a few years later.


Verisign sued ICANN! Why? What’s the full story?


> VeriSign had sued ICANN, accusing the regulatory group that oversees the Internet's technical infrastructure of overstepping its contractual authority and dragging its feet on allowing VeriSign to offer new services such as a wait-list service and internationalized domain names. In the lawsuit, VeriSign claimed that ICANN stepped outside its charter by delaying the introduction of new VeriSign services, including its Site Finder service, which redirects requests for nonexistent Web addresses, and its ConsoliDate service, which manages multiple domains. VeriSign claimed that ICANN cost the company money because of its tactics

Details here:

https://www.computerworld.com/article/2560283/icann-revises-...


I love how they characterize blatant abuse as "services".


Here is the ICANN archive of documents related to SiteFinder.

https://archive.icann.org/en/topics/wildcard-history.html

In the end it resulted in what ICANN now calls the RSEP process.

https://www.icann.org/resources/pages/rsep-2014-02-19-en


TLD-insensitive request sounds like a nice feature: when you type cia.org and is redirected to cia.gov


Except when you want to find out more about the x.org X11 server and get redirected to Twitter (x.com).


Was there ever a time someone who wanted to find out more about xorg x11 knew (without already knowing about it) to just type "x.org" instead of "x11.org" or "xorg.org" or "xorg.com" or "xserver.org" and so on, finding non-existent responses or alternative groups who call themselves x?

Collision and needing to "divine" the domain seem inherent to having a name system. Both this and the need for a TLD-insensitive lookup were solved by making the address bar also the search bar so people can use search engines to find new things instead of DNS.


Of course, the actual solution to this minor problem turned out to be the web search engines.


Do you trust them to not send you to a phishing site?


About as much as I trust the DNS to not send me to a phishing site.


Aka, all TLDS become worthless and the guy with the most money wins?


They already serve informational purpose at best, the site belongs to a tld with certain restrictions, but why tld restrictions should be expressed in the domain name?


TLDs can be thought of as namespaces, and they have good reason to exist.

For one thing, changing how domains work like this would massively drive up the cost of any one domain to the point where hobbyists and open-source projects would be priced out. I have a few domains, none of which are duplicated across different TLDs, and each of which serves a different purpose. It would have been pretty much impossible for me to do this if everything was under one namespace.

In addition to the price aspect, it would pretty much force the system into being a much more restrictive version of the already-existing trademark system - except that there wouldn't even be any exceptions for different fields, as there is in the current trademark system. For example, one of my domains coincidentally (and unknowingly at the time I registered it over 20 years ago) collides with the name of a movie. It's in a completely different field (actually, it's a personal, non-commercial site for me and not much is public on it) but I almost certainly wouldn't have been able to get it if TLDs didn't exist.


If you have no duplicates, why can't you have them under one namespace? Namespaces are supposed to separate duplicates so that they don't conflict. Without duplicates namespaces don't serve their purpose. TLDs can be kept as a supplementary technical information, like, e.g. cname.


But there are duplicates. There are a vast number of cases where a domain name under one TLD isn't held by the same person/company who holds the same domain name under another TLD.


DNS from ISPs used to do this too, right?


Right. Telekom (germany) had this around 2010. It was called "Navigationshilfe" ("navigation aid"). At least there was an opt-out option.

Edit: seems to have been there from 2009 till 2019 (!) after users took legal actions. See https://www.golem.de/news/t-online-navigationshilfe-telekom-... (german)


This reminds me of another ISP scam. Around 2010, my ISP would occasionally inject pop-up ads into clear text HTTP pages. Apparently the solution was to call in a complaint. The customer support agent would act surprised (lol) and promise to investigate, and no ad would be delivered to the complaining customer afterwards.


This was one of the main reasons HTTPS took off. The other was Firesheep.


Yea, everyone gets mad at Google for requiring https nearly everywhere, but outside of actual hackers, the ISPs were the cause of this.

Redirecting you to an add page is terrible enough, ensuring that the ads on that page eventually served malware was the icing on the cake.


AT&T was doing this in 2021. They called it "DNS Error Assist."

I imagine they're still doing it.


They implement this at the edge, so bypassing the nameservers bypasses the silly search page and doesn't change the authoritative domain name. Verisign was changing it at the root, for everyone.


A difference of degree, not kind. This technique is the predator of the attention ecosystem, singling out the "old and weak" if typos imply weakness!

An angle no one has mentioned it how this played into googles dominance. These predators made it legitimately safer to type into a search box than a URL bar. At least for a little while.


> singling out the "old and weak" if typos imply weakness!

They don't.


Do you have an actual disagreement to communicate, or just thought-terminating dismissal, because it seems that typos would be more frequent for users with poorer attention to detail, which sounds to me like the kind of user that is more likely to fall for a scam.

So yeah, preying on users who make frequent typos would also serve to target less observant users, who have the potential be exploited more easily than the general population. AKA: typos imply an exploitable weakness

Case in point: Verisign and Telekom squatting on typo'd domains to extract revenue from exploitable users.


Having a weakness in typing isn't what "the weak" means. There's no need to be overly emotive. We can define its badness objectively.


Arcor had this around 2004 iirc.



In Romania, one of the top 5 ISPs, telekom.ro, is still doing it in present days.


One extra reason to never use the ISP DNS services. Just switch to Cloudflare and never look back.


Cloudflare has control over too much of the modern web infrastructure. 5/10 sites I visit use it and its captchas are annoying as hell.

Better use OpenDNS, etc just to have more diversity


That was what initially spurred many people to stop using their ISP's DNS resolvers.


It is the norm here in Texas. All the ISPs available to me do this.


How much do you think you'd make typo-squatting every .com and .net with parked domain ads? A billion dollars a year?


> In the weeks following the change, the site came into the top 20 most popular sites, and reached the top 10 in the aftermath of the change and surrounding controversy.

1 Billion dollars a year is a good estimate.


High margin so a $20B market cap?


With growth like that 20b would be a bargain.


Assuming you mean they get all URL typos for free, probably a lot less than days past. How many people can only ever Google things? Directly typing in a URL has to be a tiny minority of users.


Even less than you think, because even when they type in the URL it’s into a search bar which is smart enough to correct the typos.

Edit: What’s really going to bake your noodle though, is that given all the issues with various kinds of URL squatting, they’re actually safer than those of us typing in the URL directly. Let’s hear it for my Dad, cybersecurity thought leader.


Here's a recent example of typosquatting ads serving malware from the first Google search result:

https://threatresearch.ext.hp.com/adverts-mimicking-popular-...


> they’re actually safer than those of us typing in the URL directly

I heard there's something called a "bookmark" which makes it so you don't have to type in the URL though.

But that's arcane magic.


Never got into bookmarks, seemed slower to have to grab a mouse and dig around to find the bookmark I was looking for, vs ctrl-t then the first few letters.


Except that bookmarks are searchable, and in most browsers that search is a) available by hotkey, and b) offered as suggestions in the autosuggest of the address bar, so bookmarks tend to be useful to me at least.


Firefox will search for bookmarks from the address bar in exactly the way you describe.


It can be configured to only search the bookmarks, not leaking the sites you visit to the search engine.


Fun fact: Firefox adds your whole window (url/search bar and pinned bookmarks) to your tab-stop queue. So you can just press tab to select the pinned bookmarks and use arrow keys to navigate the bookmarks.

That is, unless, the site you're on is Technically Fucking Braindead and decides to intercept keystrokes. Fuck those sites though. There's an about:config to disable that behavior but a some certain sites stop working entirely without it. Google Documents... I'm looking at you...


The omnibar (do they still call it that?) searches your bookmarks and it's more reliable than hoping it doesn't age out of your history.



The whirring sound was the late, great, Jon Postel spinning in his grave.


Tangentially, what are people using for DNS+e-mail today? Was a happy gandi customer but have migrated to Gandi for DNS and protonmail for e-mail since the Gandi takeover. I am really looking for a No Bullshit™ host for these things.


I didn't knew Gandi was taken over (I was never a customer).

I recently started to use NS1, which was taken over by IBM. It's nice and fast, comes with a great API, which even their web control panel also uses. The only negative side is that it's now owned by IBM.

I use some RHEL-alike products anyway, so I didn't mind NS1 now being IBM. NS1 doesn't mention any pricing, and I'm using their free developer accounts for my personal sites. I think it will get very expensive really quick once you become enterprise customers.


Cloudflare and fastmail


If you search in the search box I think you’ll be able to find lots of recommendations.



Google search advises you to just use Google.


Ooof! Well at least now someone has read my blog (•‿•)

I was just thinking this side-thread was a bit OT. And in my book, there are worlds of difference between Google and using Algolia’s custom HN search, which is really awesome IMHO. (I use it myself multiple times per week.)

But what the heck, I’ll bite and share my recommendations.

I personally prefer this setup:

• Google Workspace for email. FastMail seems to be a very popular alternative, I’ve never tried them. There’s also Tuta and Protonmail but personally I’m not interested in E2EE email. (I much rather use something based on Double Ratchet for comms that need E2EE.)

• Cloudflare for DNS. There are lots of solid options here. E.g. AWS Route 53.

• Dynadot as my registrar. Porkbun seems to be a solid alternative.

HTH!


Any particular reason you don't use Cloudflare's 'at-cost' registrar?


I guess I use Dynadot mostly out of habit. I’ve found their service to be reliable and stable so I haven’t seen any reason to switch.


PS. I know some HN users has made book recommendation compilations based on HN threads. Would be cool if someone made the same for mail/DNS/registrar recommendations :)


Been using Migadu for emails for the past 3 years. No complaints so far.


Thanks for the advice!


ClouDNS.net + Zoho Mail. I chose ClouDNS as I have a bunch of .pl domains which at other registrars are either unsupported (Porkbun, Route53), unreasonably expensive (e.g. Hetzner Cloud wants €35/yr, ClouDNS €18/yr) or offered without WHOIS privacy screen (Dynadot). I am not excited though about having to pay extra subscription for having more than 1 DNS zone, so I'll be happy to get better recommendations.


I have moved from GoDaddy to DNSimple to Namecheap to finally, recently, Porkbun, which I love.

Fastmail has done me well for many years.


Hostpoint for DomainReg/DNS and Protonmail for Mail.

But being located in switzerland that seems to be the obvious choice, right ;) ?


Indeed. No privacy/cryptography oriented company in Switzerland was ever a joint BND/CIA operation ;)


Just don’t use Hostpoint for email (have had many bad experiences).

Stick to someone who specializes in email (ie. Gmail/Outlook/Zoho/Fastmail/etc).


Never heard of Hostpoint! Thanks for the tip!


i switched everything from Gandi to Porkbun after the takeover - have had 0 issues


This would not be possible with alternatives to DNS like namecoin and ENS.


Except those have even bigger problems, like allowing me to claim a domain that legally belongs to someone else in a way that courts can do nothing about. Or manual typosquatting (e.g. serving phishing sites from go0gle.com), again without any recourse.


What would we ever do without big brother controlling all our moves


All regulations bad, huh? Absent any, we would do a lot of serfing on the fief owned by our corporate lords.


Really surprised the HN crowd isn't more into ENS, but then again they see the word blockchain and their bias makes them turn off any ability to think rationally


Maybe make a real argument instead of attacking your own strawman?


That's fair because there's very few legitimate use cases for a blockchain.

Domain names are perfect for this purpose though, it's decentralized, secure and there's no way to claim the domain of somebody else.


>there's no way to claim the domain of somebody else.

That's a problem, surely, as it means all domain names become owned in perpetuity. That's good if you don't want governments to be able to have control, but it doesn't fit with current systems like Trademarks, but also just if there was an error or con that caused a domain to be transferred that you want to reverse. Also, over time domains become lost; which is sub-optimal in a limited namespace.


There is quite an amount of tech proposals that seem to fail to account for the fact that people actually die quite regularly.


It's my personal opinion but there's been way to much overeach from governments, registrars and various bad middlewares to make me think that resolving without interference is the #1 problem with DNS, not trademarks.

The current DNS is a bit like having HTTP everywhere, we need to upgrade to a more secure scheme.


ENS doesn't solve that problem though. Middlewares can still intercept requests, unless you download the whole blockchain locally. DoH or DoT solve the problem of snooping or changing DNS requests, or alternatively DNSSEC solves the problem of changing as well.

The part that blockchain addresses would actually get generally worse: typosquatting is already a problem, but at least most registrars are working to limit it - some more proactively than others. ENS removes any ability to prevent typosquatting at all. I have never heard anyone complain that registrars or governments are being TOO proactive in delisting typosquatters or those using domains they don't own the trademark for.


If you flip that over, then trademarks will become the #1 problem, not resolving without interference. It will be just as bad, but different.

Maybe you're thinking of trademarks as some bureaucratic intellectual property nonsense but they are practically important as well. I could own ycombinator.com in that system. What would you do about it besides changing the name of Y Combinator?


> Maybe you're thinking of trademarks as some bureaucratic intellectual property nonsense but they are practically important as well.

I do think their importance is vastly overstated yes but that's another debate.

Right now domains are also suspended for a lot of reasons, typosquatting being pretty much the only one I would describe as a valid reason. And the downsides of allowing domain suspension seems greater than the upsides.

I'd be okay changing my mind on the subject if DNS was used as intended and not as a political tool to suppress newspapers or block pirate websites.


I haven't read RFC 882 all that thoroughly, but the section of it titled "Authority and administrative control of domains" strongly suggests that DNS from its very beginning allowed the name servers to implement whatever policies they wanted over the domain(s) they have authority over, so... DNS is actually working as intended?


You can't solve social problems with technical solutions, usually. One exception, for the time being, is Tor. But that's only temporary until they ban it.

We take a free internet for granted, but in places like China, the government knows exactly what you're doing on the internet. They have to, because they route your packets, and they won't send your packets to places they aren't already spying on. No technological solution can change that, except possibly by constructing a physically parallel internet, and that stops working when...

An FSF member recommended Tor to Uighurs. An Uighur responded: "You don't understand. They aren't just monitoring my Internet. They're living in my house."


> You can't solve social problems with technical solutions, usually.

I'd argue that historically it's the opposite, most social changes came from technical changes.

It's not about making it impossible to block websites but to increase the cost of doing it further.

There's a lot of in between places from Sweden to China and raising costs of doing the bad thing globally works.


Does DoH not solve this?


I don't think so because domains can still be suspended. It does solve some of the issues though


Trademarks already have trouble on the Internet. Who gets dove.com, the chocolate company or the soap company? If I create a new Dove birdseed company and I register dove.biz will I get to keep it?


It seems like a good idea in theory but even if Chrome/Firefox/Safari decided to adopt it we'd probably have to wait 20 years before it "just works".


You can add .limo and it'll resolve


The bias is justified. I use crypto sometimes as a way to send money without linking it to my real identity, but that's all I use it for. I have no interest in how it's implemented, only that I have total privacy (at least with Monero).


How does that total privacy work?


Advanced cryptography. OTOH:

* Each transaction is signed by a "ring signature", where it's known that 1 of a set of 8 keys signed the transaction, but not which one. This is your actual input plus 7 other random inputs drawn from the same probability distribution as actual inputs.

* Double-spending is prevented by a "key image" which is something mathematically related to your actual key (not sure how that's verified without knowing which key it is). Each key image can only be used once, or it's a double-spend.

* Input amounts and output amounts have to balance using some kind of zero-knowledge proof.


When you make a transaction on the Monero network, only the actual sender and recipient know the sender, recipient, or value of the transaction. This makes tumblers unnecessary and removes that whole ethical burden. If your Monero is ethically sourced (i.e. you mined it or bought it with your own money) then you can safely spend it without being subject to any tracking.


It doesn't.


More people should use monero!




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: