We're doing tech wrong - Part 1

We're doing tech wrong - Part 1

Wasn't technology meant to relieve humanity?

There are so many facets of the tech world in 2023 that make me tremendously sad whenever they're mentioned, that it's hard to focus on one in particular for special scorn. So this is the first part in a series of articles intended to explain why technology has actually made life more stressful for the majority of people, rather than giving us the leisure time we were promised.

Email

Internet email has been around for over forty years. Back then, being able to communicate with users across seas and continents was a useful and thrilling concept. No one cared about security, authentication, message integrity or junk mail. Obviously, in the intervening years, all of these problems have been solved and now email bears no resemblance to the "Simple Mail Transfer Protocol" that started the whole thing back in the day and we no longer suffer from forged messages, spam, identity fraud or any of the other...oh...no...wait...we still use SMTP and it's exactly the same as it was back then!

But wait, I hear the voices in my head cry, we have TLS, DMARC, SPF, anti-spam technology and all of these other bolt-ons! Yes - and they have all totally failed to do anything but frustrate mail administrators. Anyone can still pretend to be anyone else, spam filters are crude and ineffective (see section "All 'A' and no 'I'"), and apparently malware continues to sail through the network security barriers like there was nothing there to stop it.

Not all of the problems are the fault of the protocol(s) - TLS, DMARC and friends would all provide a great deal more protection if people actually used them properly. But so far no big providers have had the guts to fully enable them because they know the vast majority of email messages will suddenly get blocked. Good! The vast majority of the vast majority will be spam. The rest will be from businesses who haven't got their heads around configuring this stuff. What a great opportunity for their IT departments to get a rapid and thorough education.

Another issue underlying this problem is the software people use to access their mail. So many terrible ideas have flourished and then been copied by rival providers that they are now de-facto standards. A classic example of this is the concealment of email addresses. Rather than show the email address that was used to send the message you're reading, you just see a name. To see the actual email address you need to actively request it with a click or two. No wonder we get tricked so often. Some providers (e.g. Gmail) have attempted to give visual indications of the authenticity of the sender but, again, this only works if the rest of the internet adopts it too.

Also, using people's names as email addresses is, and always has been, a remarkably stupid idea; an observation that should be staggeringly obvious to anyone who has actually met anyone else in their life. Corporate email directories are comically awful e.g.:

  • Which Jim Johnson did you want to contact? We have 350.

  • Just email me - my email address is maria_garcia744622@corporate.hell

Of all the bad habits that have become enshrined in our notion of normal practices, perhaps the worst offender is...

Attachments

Typing the subtitle above caused me to produce an involuntary sigh. Email attachments. Oh, God.

Let me start from the beginning. In the beginning there was email, and it was good. People used it to communicate. It was a free (back then) and expressive medium that allowed you to send text to anyone else with an email address. If you wanted to send something that wasn't text, such as a picture, there was already a way to do it, but the bar to entry for the File Transfer Protocol was a little higher than that of email which any idiot could, and often did, use. So a couple of methods were invented to transform arbitrary files into text so that they could be included as part of an email, and then transformed back at the other end so the recipient could view/use it. The text representation of the file was considerably larger than the original file and so it was a very inefficient way to transfer files - especially compared to FTP.

What you need to remember is that back then, most email clients were text-only and displayed on a terminal. So even if you could send a picture, to view it you'd need to export it from the email and use a different tool to view it. As people started using graphical workstations for accessing the internet, this became easier and easier to do. It wasn't long before the practice of "attaching" files to emails in this way became so common that the software made it look natural and simple, which encouraged more and more people to use it as a way to transfer files to people. All the time, underneath the facade, it was all based on the same old text-based email.

So easy was it to drag a file into an email and click "send", that people would do it with every kind of file, with no knowledge or interest in how large or ungainly the file may be. The email software obediently attempted to attach it and send it without complaint, which would frequently lead to users calling tech support to find out why they'd run out of email space and the computer had hung. At the ISP I worked for, we used to describe these people as sending each other "Africa.zip".

You don't send a piano in the mail! Don't email them either!

30 years later, we're still doing it, and absolutely nothing has changed. When you send a file via email, you make two extra-large copies of it: one for your sent-mail folder, and one for your recipient's inbox. Once they save the attachment, they have made yet another copy of it. These massive files never go away unless you explicitly delete them. And you don't.

Even worse, people use email to send around documents that multiple people are working on. This is the dumbest possible way to work on a document: you end up with hundreds of slightly different copies of a document that may or may not be the latest revision, scattered throughout several mail servers. Oh yes, and don't forget there's no end-to-end encryption on the vast majority of email, so they'll provide fascinating reading material to anyone who breaks into your poorly protected work systems.

For the love of god use collaboration tools like Google Docs or a wiki for doing this stuff. Send links not files. It could save the life of a stressed mail administrator somewhere.

As if this wasn't bad enough, attachments are frequently used to hack your network: clicking on the wrong attachment can end up causing your entire company to be subjected to a ransomware attack. How did we get here?

Excel

Unless you're dealing with statistics, sales figures, or matters financial, Excel is almost certainly the wrong tool for you. Despite the ludicrous number of features available in its repertoire, Excel is not a database, a listing app, a scientific tool, a data transfer protocol, or a graphics package. But it's so easy to put data into, isn't it? And you don't have time to find and learn a specific tool for the job you're doing, do you? You just need to get the job done, and you have Excel there, and you can use it a bit, so why not?

The answer is that it will end up damaging your organization and making people miserable. The "quick hack" will become a buggy, long-term, procedure that will cause so much pain for the vast number of future people who have to deal with it that they will curse the idiot who chose to use it in the first place: you. Don't be that idiot! Leave the technical stuff to the people who are paid to deal with it.

All "A", No "I"

That so many people, from journalists to researchers, are so captivated by the various LLMs (e.g. Chat GPT etc) seems to be yet another example of the dangers of innumeracy.

Most people don't realise that the technology underlying these "miracles" is, for the most part, at least forty years old. The only thing that's changed is the affordability and simplicity of computing power which is incomprehensible to most people. To compare the computing power of 2023 to that which was feasible in the early 80s involves many orders of magnitude, and humans aren't good at appreciating exponents.

If people could comprehend how vast and powerful the resources need to be to service a single request, they may start to understand me when I say that AI is completely dumb.

In Kubrick's 2001 we have HAL: a self-aware AI so powerful that, without specifically being programmed to do so, can beat the crew at chess, lip-read human conversations, and become paranoid. 20 years after the time HAL was predicted to come online all we have is a pathetic freakshow of computers that can generate text and images in a way that is appealing to a large chunk of the population. As if that wasn't depressing enough, to accomplish this we need a mind-blowingly vast amount of computing power - certainly too much to be put inside Discovery One. We aren't aware of this of course, because Internet.

So with all of these resources, together with all of the money, energy, and time involved, shouldn't we have something considerably more impressive than electronic cretins that can google things?

Evidently not. AI is now brilliant and will replace us all. These digital idiots are good enough to do our jobs, create our art, and even kill our enemies. With the same level of accuracy as they exhibit at their best.

That's enough for now.

Next time: "Crypto", IPv4, TikTok, and Twitter.