I really like GaAs red, orange and yellow LEDs compared to their modern phosphor-based counterparts, as well as the yellowish GaP green, as inefficient as it is. And I also like the look of 3mm/5mm indicator lights diffused by matte plastic. However, there is a way to make them more efficient: packaging 2 chips in series into that piece of plastic! This is very common for phosphor-based LEDs in lightbulbs, where rectified mains voltage is stepped down to 100-200 V DC (or capacitively dropped and then rectified) and run through a series of LEDs: for example, 36 series chips can have total voltage drop of 108 V but to save assembly cost, only 12 5050-sized components are used with three LED chips each, wired in series. Therefore, white SMD LEDs with 2-6 series chips are widely available.
However, I'm surprised I haven't seen anyone package a good old 3mm or 5mm THT LED with multiple chips of the same color, for example 2x red or green in series for a voltage drop of 3.6 V - 4.4 V. The obvious application is an indicator light for 5V circuits where less voltage will have to be dropped on a series resistor that only wastes heat. Some mains devices also use resistive or capacitive droppers to power an indicator LED: one with 2 chips in series will require half the current for the same brightness and thus a physically 2x smaller (and thus cheaper) capacitor or resistor. Alternatively, for this application, an LED with antiparallel connections can be used to increase the flicker rate from an annoying 50-60 Hz to 100-120 Hz, plus the chips protect each other from high reverse voltage, removing the need of a diode. Such LEDs already exist (WP937EYW datasheet) but the two have different colors (all combinations of red, yellow and green are available, other manufacturers make red/white or CW/WW ones but never red/red). The alternative is to make them from SMD ones but the devices using mains indicator lamps often don't even have a circuit board and having the light-emitting chips in a package with the lens is convenient for them.
Not to mention, Joplin allows more than structured text (native embedded images/media, even more kinds of content with plugins) and the text-only
joplin.ai.chat()will definitely "need" to be supplemented with multi-modal AI support.As for new techniques, I imagine the declining AI subsidies will raise the popularity of context reuse (reducing token use by not having the LLM re-read the whole notebook; is that what it's called? I don't remember much from videos/articles about AI trends because I don't use it). This will require a huge framework for managing data structures distilling every cached context window... And I think non-LLM functions will be added too to save tokens, for example an LLM may want to know "is there an unfinished TODO?" without reading all of them.