You still likely need a human bypass. "I was injured or disfigured so my face no longer anything for you to compare to. I do not now, and never shall again resemble the image in the system."

Similar things happen when you require fingerprints or retinal scans. "Well, I don't have a right hand anymore. It's in some hospital's incinerator."

And not having an override or bypass is a software issue. Designing a system that doesn't permit exceptions or overrides is bad software design.

Yeah, absolutely horrible situation to put people in, let bugs stay unfixed that leads people to being stranded in foreign countries with no clear resolution in hours, and on top of that neither parties seem to be interested in just fuzzing up and fixing the fucking problem.

> Monique Hawkins, the acting chief executive of the immigration campaign group the3million, said it had seen previous cases similar to Webb’s, adding that it highlighted vulnerabilities in the system and it was “crucial that people have a genuinely stable means by which to prove their status”.

> She told Politico the eVisa rollout had been “rushed”, adding: “We regularly see people not only denied boarding, but also compensation, as carriers and the Home Office play a blame game shifting responsibility between each other.”

This is also a big problem at companies.

Previous company had full payments/accounts team, a few QAs, a couple accountants and SRE working on the automation of a specific payment edge case that was normally handled through support, and was taking 5-10 mins to handle per case. Whole thing took four-five months.

In the first year, less than 100 people used it, out of a few million. It was B2C, so after taxes, that amount was maybe enough to cover a couple days of lunch for the people who worked on it.

I mostly agree but I think the bar for the software still has to be unusually high when the failure mode is "you are not allowed to enter the country you legally live in"... The human fallback is essential, but it shouldn't become the thing that makes a fundamentally unreliable system acceptable

> Whoever built this didn't test it on the ground.

God this happens all the time. Software written by people that don't use the software themselves. It has happened to me so many times, I come across a piece of software with weird bugs or interactions that would be immediately fixed if the developer just spends 5 minutes using the software like an end user would. Dogfood all the things!

I bet you 100% it was built not by in house team but farmed out to an agency for £££££ millions. They will have been saying in their case studies some absolute nonsense about how they saved the government millions rather than burned loads of money too.

Also for years prior it was explicitly said share code were _not_ to be used for travels and that airlines were not allowed to ask for them.

And now that they are the airlines do not have the system to check them so you have to print the webpage which also explicitly said should _not_ be used instead of the share code...

This is the classic difference between testing whether the software works and testing whether the system works

So did we, although we were in the end able to persuade them to let us board the flight by showing a screenshot of my friend's eVisa status from the UK government website (so secure!)

To be fair to the airlines here, if they get it wrong then the UK government fines the airline. This is why they are pretty risk averse, and also why I'm surprised that the screenshot worked.

How did they roll it out embassy without having at least some old systemd or a human in the loop backup working?

I would say what we do care about is the number of people who were not able to use the system successfully through no fault of their own. Even 0.001% failure rate is too high when is affects more than a few people per year. Especially so when there is no easy way to rectify the problem.

> Do they think just throwing out a large number without anything to compare it to is somehow proving something?

Welcome to the last 20 years of UK public policy communications. Politicians know that people do not question big numbers or have the skills to put them in context.

I remember the May government shouting that "more children go to schools rated good+ by Ofsted than ever before". Me and you would say, well, there's more children alive than ever before, so it's hardly an achievement.

It only launched in the middle of April (edit: middle of February was the switchover at the border where no other fallback would have been possible)

It sounds to me like this number indicates broad success.

The rough estimate seems to be that only about two per cent of border crossings use an eVisa anyway (British passport holders don't and non-immigrating visitors don't).

Even over summer that would probably only mean maybe 2.5 million border crossings, so the other uses are not at the border — they will instead be employment/landlord/benefit/healthcare checks using the online systems.

So for example, renting a new property, starting a new job, applying for a new benefit, or needing non-emergency healthcare etc.

Without the total number of attempts, number of failures and ideally how those failures are distributed, it tells us almost nothing

Look, I'm robbing you but you have to consider that 99.9% of people never get robbed so really it's not something anyone should care about.

Why do you have to ask? Because you don't have easy access to the data. Whether or not your nation has a culture of honesty, big scale communication is always going to be dumbed down.

Actually this was developed by the government, in a meaningful sense, but it seems like it's Entrust (more or less inevitably) and Accenture who are the outsourced parts.

So basically both US multinationals (Accenture have a Dublin HQ as a fig leaf)

Would it make any difference? Fujitsu wrote software so buggy that innocent people got fined and put in prison. They still get contracts even today.

> My favorite part of the UK’s eVisa was taking a portrait photo of my 1.5y old.

I have had to take photos of each of my two kids at six months old for passports. That was not a lot of fun (though it's nice to look at the baby photo for the older one now).

The problem is there used to be a separate physical card (BRP), but the Home Office got rid of it and moved to this all digital system. In other words there is no backup method available to ground staff.

I wonder how it's possible they didn't think about this problem. Considering everyone involved in building this feature, not a single person thought about the possibility of twins or simply people who are very alike? Also, can't people open multiple accounts?

One kid needs a goatie: https://external-content.duckduckgo.com/iu/?u=https%3A%2F%2F...

What is strange is that even the best face recognition algorithms have false positive. So surely they must have planned for two unrelated customers resolving to the same embeddings.

Anyone have any idea how this can happen? Surely the boys have different first names, so how can the system mix them up?

The Home Office does manage to be both cruel and incompetent, which is the worst possible combination.

The story is about the British system not the EU one

To be fair, the rules are insane. There are several different ways in which you might be or become a British citizen, and there are five other kinds of British (non-citizen) nationality. Irish citizens are in a category of their own. And then there's rules for other EU citizens and per-country rules for everyone else.