Many techs aren't irrelevant, but do so little that they are only worth picking up when they are very cheap compared to the current levels of bread-and-butter techs. Most egregrious example imo would be reactor power multipliers. Even in pure beam ships, the weight savings are quite modest and there's the trade-off of higher explosion chance.
Less severe, but Sensor Strength should trail Sensitivity (comparable increase in range, but results in more expensive and noisier sensors and no use for passives).
AFAIK, Electronic Hardening isn't useful against current AI designs.
Some techs are optional and only worth using if you use enough. Good armour > bad shields, unless you know exactly what you are doing. ECM/ECCM tech may need a little love before the components are worth their weight (overengineered fire controls > underdeveloped ECCM). You generally don't need redundant beam techs, some individual lines there stand out too: imo railgun calibre quickly stops being worth researching, too throttled by capacitor tech. Currently, something similar applies to particle beams but Lances will change that. Some Missile Launcher and Magazine techs are not very important if you decide to rely on box launchers instead. The entire genetics line can be ignored most of the time, especially the aspects that can be remedied by terraforming.
Some things have their use, but are easy to misuse. Stealth tech on offensive ships is usually wasteful, just split the package on FACs/Fighters. Huge sensors can't be split up on multiple ships though, so while expensive stealth tech does give a relevant option for high-end recon ships.
One example where I've often seen poor prioritising is in power multiplier or fuel efficiency vs. engine concept.
Higher engine concept may give better power than maximum power multiplier tech, without the drawback of increased fuel consumption.
Higher engine concept and correspondingly lower multiplier may save more fuel than better fuel efficiency tech, and in the case of <1.0 multiplier will result in cheaper and less manpower-intensive engines as well.
Increased fuel production generally scales worse than measures to limit fuel consumption; in the upcoming version we have 2 additional logistics tech to exacerbate this one of which directly goes into this: reducing fuel consumption by 20% is much better than increasing fuel production and refueling rate by 25% - you need less weight in fuel tanks and consume less Sorium.