Build an amazing app/platform/service. Host it on whatever tld is cheaper/convenient/some_other_criteria. No one will care what the extension is - especially with current and upcoming consumption models (mobile apps, vr/ar, and so on).
Until you try to send emails from your random TLD domain and they all end up in spam.
Or when your shitty TLD has a NIC-level DNS outage. (.club had one a few weeks ago for instance, experienced it firsthand.) Or no outage, but resolves slowly.
Project: oefrha-super-app.${something}
One owns the other. Useful for running multiple experiments at once, though shifting to a different structure is better (imho) if the experiment is succesful.
Only on HN. In the real world, it matters quite a lot.
I tried to get a healthcare company to pick up its .health equivalent. The marketing department included the possibility of using the domain in its annual customer survey. The result was that a very large number of patients believed it to be a spam or scam domain.
So the tld matters a bit still.
There's also some TLDs that are well known for web spam, phishing, or email spam, sometimes because of free or cheap subdomains. Those might have issues getting indexed in Google, or issues with SMTP delivery. Google once deindexed the entirety of *.co.cc, for example.
It makes some sense, .xyz are so cheap that I've bought mines for $1 so just buying a domain every day isn't outside the realms of possbility for spammers. Personally, I just use my .xyz domains for internal stuff where I still want things connected to the internet and available.
New gTLDs - The registry can pretty much arbitrarily set prices in the future
ccTLDs - Politics might make you have to change domain (e.g. UK owners of .eu)
Pick your poison, I tend to stick with the original TLDs and my own country's ccTLD
When it comes to pricing, though, you never know what will happen down the road. I honestly don't know who I'd like to trust more: Verisign or the government of Montenegro?
Since .me is more targetet at / suitable for individuals rather than a hipster startup domain like .io, the prices will hopefully remain lower. Or at least that's what I like to tell myself.
Anyway, non-objective reasons I can think of:
Lots of people don't know anything but com exists.
Com is definitely still dominant, so someone sitting on yourthing.com when yourthing.biz.no is your real site will probably be getting some chunk of your traffic. Imagine if Amazon.co.uk was the real Amazon and Amazon.com was like, a news website dedicated to posting all the bad stuff Amazon does. Tbh that'd be nice for society but you can probably understand it would be undesirable for a company.
.io is one of the only ones that's managed to kinda read com-like status, and it's only for a niche (though a lucrative niche!) audience.
This is actually an interesting example, because Amazon.com is a US/Canada specific site, while Amazon.co.uk is the UK/Ireland site (though I'd imagine they'll eventually either set up an Amazon.ie or change the redirect to Amazon.de with the tax/import brexit implications of that). So while all domains are Amazon owned, I wouldn't say Amazon has .com primacy over regional domains.
Ditto for google search, though the rest of their products are on google.com for me (I think gmail used to be googlemail.co.uk in the UK and googlemail.de in Germany because of local companies who had the gmail name first, until Google bought out the names from the local owners).
All the web 2.0 or newer companies are .com rather than regionalised though.
As for reliability, there have been a few ".io" outages and I remember ".club" going down a few years ago but DNS makes complete TLD outages pretty rare.
I'd say the biggest reason to avoid certain TLDs is the human factor. Humans are used to typing ".com" or ".org" or ".net". I don't have any data to back that up though, so it probably isn't a satisfying answer!
[1] https://blog.nameshield.com/blog/2020/04/30/choosing-the-rig...