In July of this year, I turned 40. It’s also happens to be the 20th year this blog has been in operation.
It seems that both of these events are good enough reasons to take stock of the last 20 years.
One blog post isn’t long enough to do a comprehensive review of the last 20 years of technical innovations. I just work through what I thought were the most interesting developments over that period and my reflections from my vantage point as a senior leader in a technology company.
If you told me 20 years ago that in 2023 we’d have an AI strong enough to pass the Turing Test I’d have been astonished. I probably would have thought this capability would have arrived in roughly 2100. It seemed, at the time, totally unthinkable that we’d have something like ChatGPT in the 2020s.
What was even more surprising is a robot that puts my clothes away, makes me coffee, and helps out about the house still feels decades away. If you’d asked me 20 years ago which would arrive first, I’d have said the housekeeping robot.
Instead we have a chatbot that can write me a song or explain quantum mechanics and is used by millions of people every day for free.
I think for a lot of us, it seemed unthinkable even five years ago.
I used to watch a show called “Beyond 2000”. There was was an episode where computers were used to drive cars around a track. It talked about how in the future we’d have self-driving cars.
At the time it looked really far away, bordering on science fiction. At least 20-30 years away. It still feels some distance away, even though definite progress has been made.
It is obvious to anyone that we do not yet live in a world of iRobot with autonomous cars whizzing about on highways at high speed.
There first tentative steps are being taken in American cities, as this article describes. However, these taxis are relatively slow, get easily confused and only work in cities with relatively mild weather.
Over time they will get better and gradually become more capable. That said, I don’t expect this to be a sudden tidal wave. I think even at 2030, we’ll still be in the beta-testing phase with this technology.
It comes back to the classic AI problem given above. It appears easier to train computers’ to mimic creativity than programming them to put my clothes away.
Back in the early 2000s, if you wanted to have a presence on the Web you needed to have your own website. You largely had to design and build that website yourself. This was what I did, building this blog, and what all of my friends did too.
We all had web pages specially built, with minimum tool support, that were our little corner of the web.
I wrote this post 17 years ago about the rise of MySpace. At the time, I felt that MySpace was removing the individuality expressed by these home-grown websites. My criticism was quite sharp, declaring:
Myspace isn’t the answer, it’s the question and the answer is no.
In truth, MySpace fostered the same sort of creativity as the home-grown website but made it more accessible to people of lower technical skill. MySpace came and went with breathtaking speed. What replaced it was a set of much nastier alternatives.
These alternatives turned that diversity and sort of craziness of the early web in to a polished advert serving machine.
The algorithms that suggest the content aren’t really AI, just weak machine-learning, but they still had a very powerful effect. Content suggestion algorithms as simple as just recommending content from people who consumed the same sorts of material as you, creates strong echo-chambers. They’re so strong, in fact, that it’s a literal threat to democracy.
These systems are not optimised to make happy people, they just have a goal to maximise engagement. Angry people tend to be quite engaged. The net effect is that the systems are making humanity more angry.
This vision of the future was hard to appreciate 20 years ago.
These systems had the side effect of killing the home-grown website. People still blog today, of course they do, but they’re search engine optimised and many of them are slaves of that engagement machine I talked about above.
One by one, the websites of my friends dropped off the web. Now available only in the Internet Archive.
This site exists really as an act of rebellion - to jealously preserve this corner of the web with its 2003 colour scheme - as a monument to an earlier, better natured time.
I could see back in 2003, that mobile phones were the future. They’d already been largely adopted by everyone over a few years either side of the turn of the century. When I finished high-school in 1999, very few people had phones. By the end of college in 2001, everyone had a mobile phone.
Initially the phones were relatively basic focusing on providing mobile phone calls and the text messaging. Internet connectivity existed and basic web pages could be served using a now extinct technology called WAP.
Back then the really nerdy kids had PDAs like Palm Pilots or the clunky Microsoft PDAs, with a stylus. I think back then, the smart-phone revolution was anticipated and even expected, but the exact timing of mass adoption was uncertain. Most technical people felt that eventually these devices would converge with phones to give us some sort of handheld mobile computer.
The iPhone happened and the rest is history.
I think what surprised me most was the sheer strength of the hardware and the number of features that would be built in to your mobile pocket computer.
My current phone, the Samsung Galaxy A21, has:
This is significantly better than the desktop PC I had in 2008 and it costs only £135. I probably would have been surprised by the capability at that price point if you’d shown me the device in 2003.
Perhaps the most interesting part of that revolution was the fact that Microsoft and Palm completely missed it. That was quite shocking, given their position in the market at the end of the 90s. It shows how important timing is in business.
MSN Messenger was the way I communicated with my friends 20 years ago. I used it every day to talk to everyone in my peer group.
It finally died in 2013 and it was a moment of sadness, for me personally, but I think for many other people too.
However, the capability never really went away. With the rise of the smartphone, WhatsApp and Telegram basically fulfil the same role today. Except now it wasn’t just computer enthusiasts using it. My parents use WhatsApp now on a daily basis. They never used Messenger.
Back when I started my career the term “Software as a Service” (SaaS) did not exist as an industry term.
When I started working my first job at EWC Ltd in 2000, the first piece of commercial software I ever wrote was a basic web application.
EWC was in the business of taking waste from their clients’ site and transporting it efficiently to landfill or recycling. The application was written in Classic ASP and the purpose of it was to allow our clients see how much waste they had uplifted in a given time period. The users could login via the internet and see these reports.
This wasn’t strictly software as a service as they didn’t pay a subscription to use this application. However, when I moved to MPP Global, in 2004, I started there by writing software that was API led, delivered via the internet and the clients paid for access to those APIs on a subscription basis.
In many respects we were doing SaaS before it was cool. This business model ended up exploding and it’s practically impossible for businesses operating today to avoid SaaS.
Some people made a lot of money on BitCoin. Most people lost money. With NFTs, the situation was worse. Almost everyone lost money.
As someone who took cryptography really seriously 20 years ago, it felt almost immediately to me like a solution in search of a problem.
With ChatGPT entering the market at the end of last year, the hype around AI seemed to fatally wound the idea of a Web 3.0 built around NFTs.
It’s probably too early to conclusively state that as a historical perspective. We’ll see over the coming years how well this paragraph ages.
We’re still only a few years past the Covid pandemic and perhaps not yet far enough way from the events to fully judge its impact.
From our current vantage point in 2023, it appears that it radically changed the working landscape. The last full day I had in the office - where everyone who worked there attended - was in mid-March 2020. Since that day, we have not had a full day in the office and it looks like that for the foreseeable future it is likely to stay that way.
At the time of writing, there is a concerted push amongst many businesses to try and return people to the office. It remains to be seen how successful this will be.
My prediction is that this push will largely fail. We now have the technology to work from home and for many office workers it’s a significant improvement in quality of life.
I think the pandemic occurred at special point in history. 15 years earlier and lock-down would have been been significantly harder on the economy. It would have been so hard that I think politicians would have avoided it. 15 years later, and mRNA vaccine technology will shorten the vaccine development times by so much that no lock-down would have been necessary.
The fact that it occurred when it did led to a significant acceleration in the use of homeworking. Had it occurred at another time, homeworking might not have happened to the same extent.
I wrote this article to create a snapshot of my thinking on technical progress in the past two decades.
The last 20 years has seen a really interesting set of technical developments. It’ll be interesting to see where some of these end-up. At the time of writing, AI looks incredibly exciting and incredibly daunting at the same time.
Will we see artificial super-intelligence in my life-time?
It looks likely from the armchair in 2023 but AI has often been plagued by periods of excessive hype, quiet incremental progress and periods of AI winters - where not a lot happens. AI will only get better over the next 20 years and I think the economy will have significant changes when we look back in 2043.
I think over the next 20 years, we’ll definitely see the arrival of robotaxis. I think adoption will be slow but gradually we’ll see city by city adoption until some critical mass is reached.
In many respects, due the developments in many technologies, the next 20 years might have more fundamental advances than we saw in the first 20 years of this century.
This is terrifying and exhilarating in equal measure.