The difference between a phone and its predecessor was once substantial. It has become modest enough that many people skip several generations without noticing much.
Several things caused this and they're mostly not failures.
The obvious problems got solved
Early smartphones had glaring deficiencies. Screens were low resolution, processors struggled with ordinary tasks, cameras were poor in anything but bright light, battery life was inadequate, storage was constrained.
Each generation addressed a genuine limitation, and the improvement was immediately apparent.
Those problems have largely been solved to the point of adequacy. A screen with sufficient resolution that individual pixels aren't visible cannot be meaningfully improved by more pixels. A processor fast enough for the tasks people perform is fast enough.
Diminishing returns follow from having addressed the constraints, which is what progress looks like when it succeeds.
Silicon scaling slowed
The underlying technical driver.
Improvements in processor performance historically came substantially from manufacturing advances producing smaller transistors. Those advances have become more difficult, more expensive and slower.
Performance gains per generation have narrowed, and the cost per generation has risen considerably, which affects both what's achievable and what it costs.
Manufacturers have responded by adding specialised processing units — for image processing, for machine learning — which deliver improvements in specific tasks rather than general speed.
Physical constraints
Several dimensions have reached practical limits.
Size. Phones cannot get much larger and remain pocketable, or much smaller and remain usable.
Battery. Energy density improvements in lithium cells have been incremental. Battery life improvements now come mostly from efficiency rather than capacity.
Camera optics. Sensor size is constrained by thickness. Improvements come from computation, which has its own diminishing returns.
Connectors and ports. Largely settled.
The economic side
Replacement cycles have lengthened substantially. People keep phones considerably longer than they did.
Several factors: devices last longer, the improvement per generation is smaller, prices have risen, and the disappearance of subsidised contracts made the cost more visible.
Longer replacement cycles reduce manufacturer revenue, which has produced responses — higher prices, more model tiers, more services revenue, and trade-in programmes designed to shorten cycles.
Where improvement continues
Not nothing, and the areas are specific.
Computational photography. Still improving noticeably, particularly in difficult conditions.
On-device machine learning. Enabling features that previously required a server, with privacy and latency benefits.
Efficiency. Better battery life from the same capacity.
Connectivity. Successive wireless standards, though the practical benefit depends on infrastructure.
Software support duration. Genuinely improving, with several manufacturers extending commitments considerably. This may be the most consequential recent development.
What to optimise for instead
Given that performance differences have narrowed, the factors that determine whether a phone remains good are different from those the marketing emphasises.
Software support length. Determines how long the device remains secure and usable. A phone with several more years of updates is a better purchase than a marginally faster one without them.
Battery replaceability. The most common reason a functional phone is replaced. A device where the battery can be replaced economically has a substantially longer usable life.
Repairability. Availability of parts and cost of common repairs.
Bloatware and advertising. Affects daily experience considerably and rarely appears in specifications.
Durability. Water resistance and glass quality affect whether the device survives.
These are the differentiators now, and they're the ones reviews weight least.
Buying older deliberately
The practical consequence, and it's a reasonable strategy.
A flagship from two or three generations back frequently performs adequately for ordinary use, at a fraction of the price, with the main consideration being how much software support remains.
The mid-range has also improved considerably. The gap between mid-range and flagship in everyday use is much narrower than the price difference suggests, with the differences concentrated in camera performance and sustained load.
For most people, a good mid-range device with long software support is a better purchase than a flagship, and this has become more true each year as the plateau extends.
What would count as a real change
Worth asking, since the plateau is in the current form factor rather than in the category.
Genuinely different form factors — folding and rolling displays — are the most visible attempt, and they remain expensive with durability questions and unclear benefit for most people.
Substantially better battery technology would matter more than anything else, and progress in energy density has been incremental for a long time despite considerable investment.
And a shift in how phones are used — towards being a hub for other devices rather than the primary screen — would change the calculation, though attempts at this have repeatedly failed to find an audience.
Absent one of those, incremental improvement in a settled form factor is what the next several years look like.
One practical consequence of the plateau: the second-hand market has become genuinely good. Devices that are three or four years old perform adequately, and prices have fallen considerably relative to new.
The things to check are the same ones that matter for a new purchase — remaining software support, battery health, and whether the manufacturer has a history of maintaining older models. A refurbished device with two years of updates left is frequently a better proposition than a new budget phone with a shorter support commitment.