alt Hacker News98% Isn't Much
Comments
This applies to AI too.
Your classifier might be 98% accurate and it may sound like a lot.
But if it sits inside a car, making thousands of decisions during every trip then you may be in deep trouble.
It reminds me of when people argued against covid restrictions saying that the virus had a 99% survival rate. A disease with a 1% infection fatality rate is a terrible disease!
Full disclosure: I also argued against covid restrictions, but not with this terrible argument.
> Truly robust engineering isn’t about what works for most; it’s about gracefully handling the edge cases.
How do you justify this when you factor in cost and time?
> 98% is great for exceptionally good things, like dramatically increasing someone’s quality of life, but very low for basic expectations, like a baby surviving a babysitter taking care of them.
this is your brain on data science. so absurd that i laughed out loud when i read "like a baby surviving a babysitter taking care of them" like what is that phrase doing in this sentence and argument
Jesse, we need to refactor the edge cases Jesse
Feels like it's 2006 again and we're talking about IE vs Firefox.
I like to think it depends on what the actual topic is. Even the article's examples reinforce this.
98% market share? Amazing. 98% browser support? There are 15 billion screens in the world. 2% of that is 300 million. Hardly a number we can ignore. Edge cases for those 2% should be considered and implemented
Relavent XKCD comic: https://xkcd.com/325/
Hover text: You can do this one in every 30 times and still have 97% positive feedback.
I think this single fact is a major source of enshittification in large software products, especially in the era of ML/AI. If your quality is 99%, it sounds like "you have solved your task", but in reality there is a long tail that over time affects nearly every customer.
I've seen this so many times. 99% of search results are good (so within 100 queries you'll hit at least one bad result with p≈0.63), 99% of dashboard panes load normally (so a dashboard with 20 panes is broken in nearly 1 in 5 loads), and so on. If your LLM gets 99% of tool calls right, nearly every session will contain a malformed tool call.
Probabilities are hard for humans, probably.
I'm fine if you want to hand over 98% of wealth in the world, its a lot for me
Nice in theory, in practice I remember having to support Internet Explorer about 4 years ago. Hard to justify the investment sometimes, at least polyfills gave use some sanity back. The only reason to do it was: Rich old enterprise customer who can't install chrome due to policies created by Dinosaurs.
Websites are surprisingly hard to maintain long term, specially for a broad audience of devices. Developer Experience can lead to better UX, the easier it is to build/maintain, the more likely we're to do it.
Given how bad AI is at design plus all the unstoppable slop train, I expect websites to become much, much worse.
1% failure rate of a hundred might be acceptable. 1% failure rate of a million is not.
Isn't that a named law?
It's just mathematical expectation.
Don't look at the simple probability - look at probability * value.
I agree the general premise but do not agree when it comes to browser support.
I feel like we should be building for the 98% or even 95% and force the remaining to upgrade their browsers. I've built for the IE6 - IE11 era for a painful and long time. I do not give a shit if you want to use a 3 year old browser at this point. Go look at a blank screen.
Covering for the 2% is often not a sound first order business decision. There's certainly higher order benefits. A lot of accessibility features are just plain useful for anyone. And I think companies like Apple generally get that it's a kind of loss leader. But this article makes dishonest TV Shopping Network style arguments like pointing at 150 million as if that was ever an addressable market.
I think either you argue for regulation, or you argue more honestly: asserting that the extra cost will likely never directly pay for itself, but it is some of the secret sauce that can a good product into a great one.
This analogy is bad: Nobody is going to die or get food poisoning because their old browser doesn't work on a website.
A better analogy would be a restaurant deciding not to cater to the 1% of the US population that have celiac disease (cannot eat gluten), or the 2% that have issues with dairy.
In today's world of AI it's fairly easy to make your site compatible with every version of internet explorer ever.
Just tell the AI to do it. It'll find a way. The maintenance burden for you will be minimal because the AI can keep the legacy compatibility bits in sync.
Can you imagine a venue refusing entry to former clients 2% of the time just because they’ve “improved their experience”?
Sure, it's called a 'dress code'.
I used to try to make the point with non-tech people using the salesman analogy: If you were a salesperson who worked inbound calls from potential customers, would you be willing to handle 1 out of every 50 calls by picking up the phone, yelling "fuck you" into it, and hanging up? That's pretty much what you're doing to your customers when your software works for 98% of them.
This whole article is a categorical error. Whether something is good or not entirely depends on the frame of reference and the context. You can argue endlessly by shifting the topic that 98% is used on. I guess that's what people are doing here.
I wonder how much traffic from bots is skewing OPs nested CSS calculations
I am not exactly sure what is the article trying to point out
Or, phrased another way: there's a reason why we consider basic availability in nines and 2 nines is still considered pretty bad. 99% uptime means being down over 7 hours each month.
while true, the people who will read this and then think twice about implementing and applying things are exactly the people who already doing too much thinking
I blame it in big part on the WebDX community group, their absolutely useless "Baseline" guidelines, and on them allowing Apple to be part of that group and make decisions on what features are "ready" to use whilst being behind the only non-evergreen browser in 2026.
The "baseline" means nothing. The percentage in caniuse means nothing. The only number that matters is the number of Safari users stuck using a no longer supported Apple device that access your website. Of course Apple makes sure to hide usage stats of older devices.
Everyone complains about only having three browser engines out there, but I'll be happy to go down to two if that means freeing the world from Safari.
60% of the time, it works every time
I mean 2% have their javascript turned off (either on purpose or caused by failing extensions). 5% are behind corporate proxies that block your domain. Are you going to host the site on substack also so those 5% can access it?
When I read the title, I instantly thought about these crappy Ai api status pages. People somehow are forced to accept that 98% uptime is alright.
The author seems to equivocate by comparing completely different domains.
Whether 98% is acceptable, it depends on the cost of failure, not the percentage itself.
I had this argument with people working on VR headsets, where a physical parameter was designed to cover the 5th to 95th percentile. I had to point out that flat-out excluding 10% of the population is a pretty crappy starting point...
You can make it so those 2% are dealing with ugly -but functional- layouts
it that reduces development/maintenance cost by a lot, that's not a terrible deal.
Venues did kick out a lot more than 2% of their existing customers until they upgraded their bloodstream. Website visitors can upgrade their browser.
It depends on your requirements. The term that should have been mentioned is https://en.wikipedia.org/wiki/High_availability
Off topic (and at risk of being downvoted), I don't think I'll ever get a better chance to insist here that
"99 and a half won't do"
Holy Disciples
Trying to Make a Hundred
It's almost as if context matters for random numbers. A 98% success rate for a parachute is criminal, but if I could achieve 98% of my goals, I couldn't be happier.
Design bloggers are about to reinvent the concept of availability https://en.wikipedia.org/wiki/High_availability#Percentage_c...
Software standards are way too low these days. If you can't do at least 5 9s in everything you ship get out of the industry and humanity will be better off.
If for 2% of users a webpage will not look as awesome as intended (it's not guaranteed that it will be broken), that's ok. It's not poisoning - it's a 98% chance of getting a top mark.
I couldn't agree more. BTW, 98% of US users have JavaScript enabled in 2025.
Depends on the context
I thought this would be about AI slop.
>> But a restaurant where clients don’t get of food poisoning 98% of time is getting people sick on a monthly (or even weekly) basis.
Objectively, I think it's impossible to work in the food industry and avoid food poisoning 100% of the time. One of the reasons I never attempted several of my food industry business ideas. I'm certain they would be at least profitable enough to keep going, would be rather trivial to access EU subsidy money in the €50k, but the amount of regulations and inspections terrifies me. And I'm sure at some point, some salmonella or what else would slip through and don't wanna deal with the consequences.
Easier with programming computers since a "bug" won't make people expell waste simultaneously through both incoming and outgoing food orifices, like it happend to me last time I ordered sarmale from a local restaurant. Like in the food industry a "bug" is literally that.
[dead]
[dead]
[dead]
[dead]
That's why things never get better. I mean, i'm all out for retrocompatibility, but if removing something makes my experience much better and unfortunately the thing stops working for 2% of the people with outdated devices/browsers, it's not that sad, but the tradeoffs need to be measured.
I've been in the two ends of this situation, in the 2% with older iPhones and Android devices, and in the 98% with new devices. The 2% cannot hold a tyranny over the absolute majority, and vice versa. Everything must be evaluated on a case-by-case basis.
We shouldn't go out of our way to support IE11 anymore, sorry