12 ms·
Keep in mind that on Android, the default browser (Chrome) doesn't support extensions, which excludes adblock (or at least makes it considerably less convenient
by dredmorbius 2y ago
Keep in mind that on Android, the default browser (Chrome) doesn't support extensions, which excludes adblock (or at least makes it considerably less convenient, e.g., setting up PiHole or equivalent on the device).
And Google Search's past* behaviour was far less toxic.
NB: I've relied all but entirely on DDG search for over a decade. I'm increasingly finding FastGPT from Kagi useful for search-framed-as-question, though the answers have to be carefully vetted. One advantage of Kagi's LLM is that it provides references for its answers making vetting far more straightforward than with, say, OpenAI's LLM.
- eviks 2y agoSure it's not very convenient, but adblockers still exist, and the author had enough time to show the full install procedure And my criticism isn't about toxicity as in "ads are bigger", but search result quality, where the level of degradation is an open question. Like with your DDG example: I've tried to switch to it permanently for all search many times, but couldn't because found many instances where Google was just better (think DDG is mostly Bing, which isn't better vs Google)
- dredmorbius 2y agoFor a long time my significant use case for Google Web Search (GWS) was date-ranged search. DDG had relative bounds (past day/week/month/year), but not a custom range. That changed a couple of years back, and has been excellent. Now it's mostly really old websites, > ~20 years. Here Google's crawl still seems more robust than DDG's / Bing's. (And yes, DDG leans heavily on Bing.) Again, I use DDG and retry failed searches via !g bang search. But that's rarely any better. FastGPT via Kagi is transformational in that I can ask questions and get answers, rather than turning up web pages. Kagi's GPT adds links which reference its sources, which is hugely more useful than ChatGPT in my experience. FastGPT still occasionally hallucinates, but its sources can be much more readily verified. ChatGPT OTOH, hallucinates sources when requested to provide them.