As with all things, not every user is the same. The raw number of visitors using each browser is rather a useless metric for a lot of sites - you should be looking which browsers convert to sales more, click through adverts more, spend the most time on the website, etc. In my experience IE users are much more likely to complete an interstitial signup form or leave a site via a banner than Chrome or Firefox users. IE may have fewer users, but those users are much more valuable if you're building an ad-supported site.
Also, if you're making a substantial site with, say, 10 million MAUs, 5% of them is 500,000 people. That's a lot. Do you really want that number of people telling their friends "The site doesn't work for me"?
If your point was "Why bother if you're building a site where the IE traffic is of no consequence?" then you should have said that. Obviously there's no point spending time and money writing code that no one benefits from, but that isn't the case for every site. Plenty of sites still need to work in IE because it will affect their income. If someone out there is writing a website whose only customer is an eccentric billionaire who spends $100m a year but insists on using Mosaic 2.0 on a dusty old 286, then they should be testing their updates on Mosaic.
Go where your customers are. If you don't have customers, do what you like. Although I'd question whether you even need a website if that's the case.
If 'a couple of illiterate smugs' represents 3% of your userbase then you only have 67 users to worry about. I don't think I'd do much optimisation for a site that small either.
I know this is HN and everybody is building the next Facebook and has humongous numbers of users. But there is also local bands, volunteer ran music venues and such in a city of 200k tops. Which is apparently unworthy of HN. Noted.
I don't believe that's the case. Most startups, even YC startups, have no aspirations to be the next Facebook with it's billions of users. They just want to do as well as possible in their market. Every startup should aspire to serve it's customers as well as it possibly can - regardless of whether that's a couple of hundred thousand people or a couple of billion people.
But that has to include the all the customers, not just 97% of them. Start with the low-hanging fruit of course, but never settle for "those customers are too much effort to bother with". That sort of negative sentiment is toxic.
Right; in some cases it's not a personal decision.
We have a set of users who are using our site in healthcare environments, on locked-down computers that are difficult/expensive, so in some cases they're stuck on IE7/IE8.
I'm pretty sure calling them "illiterate smugs" wouldn't go over well.
On the other hand, these computers don't have access to the general internet, so in this case they don't risk running afoul of site creators who decide to be "clever" and go out of their way to aggressively (or even insultingly) demand that visitors upgrade. But restricted corporate users who hit that sort of thing would be unamused.
Also, if you're making a substantial site with, say, 10 million MAUs, 5% of them is 500,000 people. That's a lot. Do you really want that number of people telling their friends "The site doesn't work for me"?