Many from 6 years ago seem prescient: https://lobste.rs/s/p9zihz/technology_predictions_what_s_next
Predictions are hard:
- The Beatles’ Decca audition
- There’s no chance that the iPhone is going to get any significant market share
So at the risk of being totally wrong, what do you think is next in your field?
Personally I’m focused on programming and entrepreneurship. Also think it’s interesting to hear from people in the different fields: hardware, devops, machine learning etc.
I’m going to be a bit optimistic (~ 2035 - 2045):
In no particular order:
OpenAI will collapse, the hype around LLMs will subside. There will be societal swing against LLMs to the point that even using them where they are useful will be difficult.
FOSS will completely die out as a culture, but most software will continue to be FOSS purely for pragmatic reasons. That said, it will become impossible to exclusively use FOSS.
“Hacking” culture (in whatever incarnation) will become incompatible with society as large. Yes, there will be holdouts, but you won’t be able to participate in society without proprietary devices and services that track you.
Vi/emacs will become impractical to use. Any new tool will come with VSCode integration and an expectation that VSCode is the only way to use the tool.
Software will become slower and software engineering will become even more byzantine and complicated. New wrappers over the existing wrappers will be developed.
Strong encryption will be made unlawful. Backdoors will be mandated in all software you have to use day to day. Anonymity will be illegal.
Europe will fall even more behind the rest of the world when it comes to technology. Europeans will continue to feel smug about it.
Civil rights will continue to be eroded in Europe. The European Union will become more authoritarian, all in the name of protecting democracy.
Wow, I feel strongly that the opposite will happen for almost all of those, except possibly 5 and 6, possibly 7.
I really wish I didn’t but I find myself agreeing with #2 and #3, at least as far as the next 15-20 years are concerned. I think that “personal computers” as we understood them in the 1990s will be slowly relegated to their Homebrew Computer Club roots – hobbyists tinkering with interesting, but ultimately significantly less capable platforms. Computing at large will increasingly move back to a “mainframe”-like model, via portable terminals which we’ll keep referring to as “phones” out of inertia, tablets, and so on. The incentives aren’t (primarily) technical in nature but economical. It’s hard to argue with the bottom line of a subscription model.
That may not be the worst thing in the long run. All the nominally useful bits of modern computing make it a little difficult to have fun with computers these days. Liberated from the bottom line constraints, personal computing platforms may ultimately become sufficiently “free” to be disruptive again. But I think we’ll see at least a generation or two of programmers before the tide begins to shift again. I’m not sure you and me will be alive to see it swing back that hard :-).
I’m cautiously optimistic about something else though: just like minicomputers and their software were, in part, a response to the enshitification of mainframe platforms, I wouldn’t put it past our generation to come up with a productive alternative to the enshitification of cloud- and subscription-based platforms. I’m not sure what that response would be; it won’t come by looking back on the good old days, that’s for sure.
With a new Cold War looming over us, I also fear the way #6 is going to unfold. I don’t know if strong encryption will be directly made unlawful; the rich and powerful who wield power by proxy (through lobbying, donations and the like) won’t be easily convinced to shove the cat back in the bag. What I suspect is more likely to happen is that stronger encryption schemes won’t be made public anymore, and existing ones will gradually get broken.
Feeble (and crumbling) privacy laws will mean we won’t learn about the latter right away. And, via the former, we’ll end up more or less in the situation we’ve been in the ’80s and early ’90s, where economic incentives (via commercial telecom, TV and so on) tugged at strong, publicly available encryption, but under tight technical lockdown. Basically, the only strong forms of encryption available to the wider public will be DRM schemes.
At the risk of getting banned from this site, I’d say they’re already impractical, but people still use it, so, I don’t see that changing =P
I get the joke, but I’d definitely say that Emacs is more practical to use today than it was in, say, 2005 to 2010. I was trying to write Java in it, then, and the tooling for things like completion was just completely unusable - hard to set up, and unreliable and slow once you did get it set up. I’m not doing Java anymore, but with comparable things like C#, LSP makes Emacs a first-class citizen.
And it will seem perfectly justified in this timeline!
The global internet won’t be global anymore. We’ll probably have regional say north north America, South America, Africa, Europe, and Asia. That would be Asia with China and Asia without China. I suspect. Also, other regions might start to develop their own great firewalls and the current global nature of the internet will no longer be easily usable without significant hacking knowledge.
There are some reasons for relative optimism, with the split looking more like «you can fool all of the people some of the time, or some of the people all of the time» — «you can reach all of the users in some region, or all of the regions for some niche enough group of users for nobody outside it to care, but not all of the users in all of the regions».
“(15 years) It is impossible to host your own server on the “public” internet unless you’re a business.”
/me looks at banner at the bottom of the page of the UK’s Online Safety Act :(
That completely misses the point (verbatim) but maybe less the spirit.
(Basically) everyone can form a company/business - even with a low amount of resources, and some people do, for some benefits. This probably means “big, profitable? business with lawyers” - which is a completely different thing - especially regarding the Online Safety Act I imagine that a hobbyist forum without ads would have much less to fear than a support forum for a one-person software shop. I’m not saying it’s safe but if there is some sort of measured enforcement, it would be like that.
A new way of building IDEs around “gaps” blows the current IDE wars wide open, making semantic editors the lingua franca of development.
The rise of programming on tablet computers means that most children start to become software literate in grade school. Non-English speakers become broadly software literate in their native languages. 100 - 1000x more people become software-literate.
The stock markets take another huge dive as it is realized that it won’t be possible to trap humans in pits of illiteracy while coaxing them pay endlessly for language models.
Please elaborate ^^
It’s already happening. Editors that have embedded gaps include Hazel, Pantograph, and Hedy. I’m just working to take this trend and bring it to the mainstream.
ohh ok you mean “gaps” in the syntax sense. Like typed holes and the like, for program completion.
The internet as we know it will die off in the next few years.
Yes, everyone will have an internet connection, and yes everyone will use it, but not “barebones”.
You will use your internet access to connect your job(s), bank(s), friend(s), family(s) VPNs, the future is behind firewalls. Two-way firewalls, as I am waiting for malitious IoT devices to carry out private data on a massive scale any moment now.
As there is only one real browser left (the other one still breathes, but can now easily be outfeatured and outimplemented, for example ff has yet no datepicker and timepicker widgets as chrom* has), THEY can now easily implement measured digestion of N commercials per hour to the point that you have to sing aloud the (verse of the) jingle, as a sign that you have learned your message by heart.
in other words: Internet is now at a point where TV was in 2001, a few months before I abandonned my TV set. Back then as part of a lottery they tried you to stick some cartboard logo to the upper left corner of your CRT with a photo film in it that only got the correct exposure if you constantly watched the correct channel (pro7 and sat1).
The number of sites I visit daily is down to five, lobste.rs is one of them. I can not remember when I last time just “browsed” the internet, as I did years ago. I rather go to my workshop now and do something in the real world.
The number of ad free writers and bloggers has diminished. Most people want to sell you something.
In the 1990s you could download teaching materials and sometime whole textbooks for free, which all was locked away after the bologna reforms. in the late 2000s you could write an email to some ph.d. at some university and chances were that you got a friendly and short answer to the point, maybe a pointer to the right literature.
try that today.
Keeping pihole and the adblockers up to date for the Internet generates efforts and expenses comparable to descrambling pay TV in an effort to be able to watch a movie without interruption back in the late 1990s.
In 2020 in the lockdowns I briefly had netflix, it lasted 6 months. watching any show or news today feels like watching Carpenter’s THEY LIVE! to me, does not matter if on my parents’ TV or on the web.
I’m in gemini sphere too. Feels nice, cosy and nerd like the early mid 90s internet, has only one error: most people there are in the second half of their lives. As if the young students of today have nothing to say.
Most things come in apps, I think most people would be down to 3-4 actually different URLs they visit on a web browser daily. But connecting to chatgpt, facebook, youtube, discord, email, music… they’re websites that have broken out of the browser, but I’d say they’re still “the web”.
I will say that I have found more and more self sustained things in the last 1-2 years. Be it videos, podcasts, live streams, newsletters, or even whole fiction books. You just have to pay a bit. They’re still individuals doing something they, and you, care about, and the quality can be incredibly high.
I prefer the people selling their work to the people selling something else, or the platform selling me and my information/attention.
They do, just elsewhere. there are many mediums, many communities full of young people who have things to say, care, and do very cool things. To some extent people are bubbled by what and who they know: going onto platforms that try to re-create a specific time period of the internet will inevitably result in people from that time period aggregating there.
That’s been my experience as well. I’m actually quite optimistic about the world that young people around me are building. They obviously don’t hang out on Gemini; that’s something people in my generation find cool because we were around for the WWW (and Gopher, to some degree; that was a bit before my time) but that doesn’t mean anything to them.
What is young for you? I will join the gemini sphere with actual interesting stuff but I feel as a young person Gemini is a submission. The smarter heads-on attack approach is push for HTTP 1.2. Gemini is a defeat, a shelter for the wounded. I want an internet where I am a soldier for the better world with higher chances. The difference here is Gemini is for people seeking refuge. I want adoption of a better today, because my tomorrow is still far.
Well, at the risk of being totally wrong… In the next five years
The AI bubble bursts. Maybe not as catastrophic a failure as the other bursts we’ve seen in decades past, since I think that the AI of today has some value, even if overhyped. But we don’t get AGI nor do we get armies of Devons doing all software engineering. This has a few ramifications:
The VCs lose their gambit big time and the only surviving big AI players are the ones who have a solid niche (I still haven’t seen any news about AI used in the adult industry, but surely it can be a big player there, even if just with chatbots) or those who pivot.
Software engineering as a whole takes a hit, as companies discover Devons will not in fact replace their juniors. Until this point, however, the market will continue to be tough, especially for new grads. When everyone finally realizes that they still need to be hiring and training up juniors, there’s going to be a shortage of senior talent. This will be additionally and especially compounded by the average ability of junior engineers decreasing due to the combination of COVID schooling and ChatGPT.
AI tooling eventually gets folded into everything. It’s useful, but not revolutionary. The seniors who get good at using it, however, become even more valuable.
The (western) world takes a hit from the various crashes, possibly enough to start the recession that has supposedly been looming for the past 5+ years, although I think if that happens there would need to be other contributing factors than just AI flopping.
And because I can dream, Refinement types reach the mainstream, like how gradual types did with Typescript (though possibly and probably in a less successful tool).
Apple will release some
VR hardwareedit: AR hardware that replaces the iPhone (i.e. it provides the same features as an iPhone but with the virtual screen size of a home cinema). This might then also replace Macbook devices (since it offers the same hardware specs and a comparable virtual screen size). No idea about the timeline; but I think this development is kinda inevitable since the demand for larger portable screens exists but building a portable Apple-quality device with a 20” screen (e.g. a foldable screen) seems technically impossible to me. But maybe I’m wrong and Apple will release a Star-Wars-like hologram instead (well, but then with better quality than a blurry blueish figure).I predict that LLM usage will continue to rise for the next five years. I think they have proven their value to at least a part of the people; and I think therefore their rise or demise mostly depends on political decisions now (which I don’t dare to predict now; so I would not rely much on this prediction.).
machine learning will make some improvements in the non-LLM areas, which will make some tasks much easier (e.g. training an ML system to do object detection/classification with very little training data and very little annotation effort). This prediction might be difficult to validate though, since this will rarely be visible to the public (I think it will mostly replace some menial jobs).
I hope that PLCs (by Siemens, B&R etc.) die out :-D well, at least are gradually replaced by a wave of more simple, more open real-time controllers that are improved versions of e.g. 3d printer controllers. I have no idea if this will happen; but seeing the sorry state of software development methodologies in this area I think someday enough devs will be sufficiently fed up with this and will build replacements that bring modern SW development approaches to the field of hardware controllers, and that thereby replace traditional PLCs the same way Linux replaced Unixes. Fingers crossed.
Rust will become more widespread, C++ will be used less for new developments. C++26, C++29 and C++32 will still be released, but I could imagine that Clang will state that they will not work on C++32 support at all.
[Comment removed by author]
I tried to write some things about AI stagnation and NixOS dominance, but it’s too hard to predict anything very useful in the current world. Companies will pursue the AI gold rush until the money runs out or a sudden disillusionment wave overwhelms them, and while NixOS is brilliant for servers there’s too much inertia and too much of a learning curve.
The most likely scenario of AI stagnation I could imagine is iPhone-ification.
The promising multi-paradigm pilot studies (like LLM for context ingestion + AlphaZero-like for getting inference checker to approve the logic part) keep not getting followed up, easy-to-use from get-go trumps possibility of developing deep user skills, stuff like LLaMA and DeepSeek makes sure that the remaining money is in superficial niceness.
Transcription to text, style adjustment, straightforward (sometimes large-block for tasks with more boilerplate than business logic) translation to code, and summarisation fully and undeniably catch up to translation (which was already there pre-LLM with DeepL).
Reasoning coliides with safety and ends up in a niche similar to physical-keyboard hand held devices today, you can get that stuff to run locally but you need to have a combination of interests and competences, and then you get a small nice boost to your workflows.