There’s a useful way to think about innovation that doesn’t get discussed enough.
A lot of what we call new has already been tried. Sometimes recently. Sometimes decades ago. The core idea shows up, someone builds it, it doesn’t quite work, and then it comes back later under slightly different conditions.
If you look at how things actually get built over time, the same categories keep reappearing. Communication, payments, transport, coordination, health. People keep taking runs at them as the world changes around them.
What tends to determine the outcome is whether the surrounding conditions support the idea at that moment.
That’s why looking at earlier attempts is more than just interesting. It’s a practical way to judge whether something might work now.
If something failed before, the useful question is what stopped it from working at the time. Once you understand that, you can look at whether those constraints still exist.
A few examples help ground this.
General Magic tried to build a handheld device with messaging, apps, and networked services in the 90s. It didn’t get traction. There was no real mobile internet, hardware was limited, and battery life made it impractical. The idea itself carried forward. The conditions caught up later.
Webvan built online grocery delivery before people were used to ordering things online. They invested heavily in infrastructure and couldn’t make the economics work. Demand wasn’t concentrated enough, logistics software wasn’t mature, and behavior hadn’t shifted yet. The same model later works because those pieces fell into place.
Digital money showed up earlier than most people realize. Early systems were technically sound but had no real distribution and limited usage. Now digital payments are a default in many places.
Tablets were shipped before they made sense. The early versions were slow and awkward. A few years later, the same concept works because the underlying components improved and there was a real ecosystem around it.
When you look across these, it comes down to a few recurring constraints. Compute, cost, infrastructure, and behavior. When one or more of these changes, ideas that didn’t work before start to look viable.
The list of areas where this applies is long.
Autonomous vehicles have been worked on for decades. Progress picked up once perception systems and compute improved.
Vertical farming has been around since the 70s. It starts to make more sense once lighting efficiency and environmental controls improve.
Voice interfaces existed long before they became usable. They needed better models and more data.
Remote work was possible in theory long before it became common. It needed reliable connectivity and a shift in how people operate day to day.
Some areas are still developing.
General-purpose robotics is still limited outside structured environments. Brain-computer interfaces are still early. Carbon capture is improving but still expensive. Fully autonomous supply chains are still not widely deployed.
Then there are cases where multiple attempts have been made and the outcome is still unclear.
Virtual reality is one of the more interesting ones.
The hardware has improved significantly. The experience is better than it used to be. It still hasn’t become part of everyday life for most people.
Some of the constraints here are technical. Some relate to comfort, motion sickness, and how people actually want to spend their time. It’s not obvious yet how broadly it fits into daily behavior.
One thing that becomes clearer when you look at these histories is that innovation cycles don’t always take decades.
Sometimes the gap between attempts is much shorter.
Airbnb is a good example of that.
Short-term rentals and home-sharing existed before Airbnb. The idea of renting out spare space was not new. What changed was the timing.
It launched into the global financial crisis, when people were more open to finding cheaper accommodation and also more willing to monetize spare rooms to generate income. Payments infrastructure was better, smartphones made discovery easier, and trust systems like reviews were more accepted.
Those conditions came together in a way that allowed it to scale quickly.
That’s an important point. You don’t always need a long gap between attempts. Sometimes the difference is a narrow window where behavior, infrastructure, and incentives line up.
Another pattern that shows up is how teams approach the problem.
Some stay focused on the underlying problem and adjust their approach over time. Others get tied to a specific method or technology.
In AI, early efforts focused on manually encoding rules. That approach didn’t scale well. More recent approaches rely on data and learning, which handles complexity differently.
In space, newer companies run more iterative programs and learn from repeated attempts rather than trying to get everything right upfront.
How flexible a team is in adjusting its approach affects whether it can keep moving when something doesn’t work.
Right now, several underlying constraints are shifting at once.
Compute is more accessible. Sensors are cheaper and more widespread. Software is being applied more directly to physical systems. There’s more capital going into industrial and hardware problems again.
When that happens, older ideas tend to come back into focus.
A useful way to look for opportunities is to start with things that didn’t work before and understand why.
If the reason they failed has changed, it’s worth taking another look.
This doesn’t guarantee anything. It does help filter for ideas that have a better chance of working in the current environment.
Most people focus on what feels new. Spending time on what has already been tried gives you a clearer view of what might actually work now.
George Payne 