> I understand the point you're trying to make here
The point was that it's a multi-dimensional space, and the OP should actually consider the product rather than automatically go all "lol, Google product, bet it gets killed".
So if you think that "some old products were discontinued" is any kind of rebuttal, I clearly didn't make my point well enough. Of course that happens! The alternative is that any sufficiently old product would automatically become immortal, which would be a ludicrous idea.
8.8.8.8 clearly has a ton of users, which already differentiates it from basically everything on your list. It's also isn't something you could just put in a maintenance mode and forget about, unlike a lot of the things on your list, both due to the scale and due to the impact if it were to stop working. Something like "Google Showtimes" would not have been a big drain on resources for most of its lifespan... When there's a measurable cost to keeping a service running, the longevity does actually signal something about the business value.
Products with no users get killed by all companies. Products with hundreds of millions of users don't get killed with one exception: to migrate the users to a different product for the same task.
And that can't really happen with DNS! They can't replace the clients, nor force the clients to upgrade, and they can't change the protocol in a way that would force some kind of a migration. Even if they end up deciding that the service needs a full rewrite, the external interface will have to stay the same.
And once you think about the specifics, it actually becomes kind of an interesting discussion to have! What are the circumstances that could lead to this service being discontinued?
A complete migration from IPv4 to IPv6 might do it: part of the value of both this and 1.1.1.1 is that these are IP addresses that people can actually remember. Their IPv6 addresses do not have that property. Not holding my breath on that one though :-P And even if that migration ever finishes, it's plausible that operating systems start including a dropdown of well-known public DNS servers as one of the configuration options.
Could they replace DNS entirely? Come up with a "QuikDNS" that starts off as proprietary, is implemented in only Chrome and Android but never replaces DNS outside of their ecosystem? Or instead of a proprietary protocol just stop supporting classic DNS and only continue supporting DNS over HTTP? I don't see the former, there's just not enough wrong with the standard protocols for that to be worth it. I could definitely imagine the latter happening at very long timescales (like, not for at least 10 years). At the point where 99% of the traffic is DNS over HTTP, maybe the cost benefit ratio stops being there for classic DNS.
> and a lot of the things they killed were well used, long lived, and popular.
I don't think the examples you posted really match that description. Most of them weren't ever popular, let alone when they were discontinued. Maybe Picasa was?