a rather general proposal
@Vlad
Hello sir. I have been using Kagi like fire since subscribing and, the more I use it, the more admiring of the effort and care put into the ecosystem. While things are not perfect or smooth as Google, I am constantly providing feedback on the hurdles I come across from the transition.
I understand and appreciate that:
- Kagi seeks to put the value to users single and foremost; not advertisers.
- Kagi, like other successful movements (like Home Assistant), provides the community with direct communication with developers and leadership (as through this platform). More importantly, direction of the project is based on a collaboration, discussion, and idea exchange between the user community and service providers. As hectic or messy as it can be at times, I believe this is the pinnacle of a healthy movement where everyone benefits.
For me, I use Kagi as:
- A neater alternative to Google, hoping to fully replace it eventually.
- A research tool for different kinds of task (this is the focus of this post).
- A community, movement, and a powerful experiment on regaining how the Internet really should be.
If Kagi succeeds at these two milestones, I am super fulfilled as a customer (and endorser):
- Replace Google fully in terms of functionality and UI.
- Supersede Google as a research tool.
The second point is a dream I had lost over 10 years ago, when I though (and tried) to have a configurable search engine that offers a bird eye view of results, allows me to cater and filter, etc. Basically, I wanted search engine as a research tool. I learned advanced search syntax of Google, tried Advanced Search page, tried many other alternatives, tried deep web engines, and eventually realized that in the end: I get to see what Google wants me to see: a micro-filtered subset of the Internet. Any other path was uneconomical and inefficient.
As of late, many posts seem to realm around search widgets. Eg:
- Date Arithmetic #10282
- Currency Converter #3412
- Country infobox #10307
- Word Origin #10332
- Spellchecker & similar words #3645
- Islamic prayer times #9982
- More powerful translation capabilities #10275
If generic web search (and AI responses) is like roads, optimized widgets are like buses, carriage tours, lemos, portable clinics, etc 🤔.
My suggestion is: Kagi can become an extremely powerful research tool and a distinguished search engine, if it extends its mission to providing superb best-in-class widgets. I hope (dream) to have in Kagi very powerful widgets and research tools that make me never want to go back. Think of things like:
- A dictionary widget with aggregated definitions from notable dictionaries, synonyms ordered by frequency, word origin graphs, translation in my desired language (eg: Arabic), word frequency graph, image snippets for nouns, icon tiles (like share buttons) to definition at different dictionaries, etc.
- Weather widget
- Currency widget like that of DuckDuckGo and Google (as exemplified in #3412)
- Prayer time widget (as described in #9982 §7)
- There are absolutely many inspiring examples at WolframAlpha for all sorts of topics. In fact, if just a 'window' of the results and visualizations provided was integrated in Web Search (just like Quick Response), we would harness great power already available.
Now widgets in Kagi already exist. What I am suggesting is more harmonizing: seeking in a more coordinated fashion to provide widgets for different categories of users, and to try to excel at making their UI of superb ergonomics, visualization, and interactivity.
More specifically, here is what I am proposing:
- A pinned thread for suggesting widgets.
Even better (for effective voting), a 'Suggest a Widget' post type with a specialized form (namely: widget description, good examples (if present), bad examples (if present), any UI specifics, suggested data sources, value rating (eg: low/medium/high), frequency of use rating (eg: low/medium/high), etc)
- Widget design has two aspects: functionality & data sources, and user interface. Developers are often focused on the former part. For widgets to be of superb UI (in terms of ergonomics, information presentation, visuals and language/notation/viz, layout, interactivity, etc), it is optimal to hire a UI/UX designer. A good designer focuses on the little details that can make 'good enough' (user adapts to widget) to excellent (widget provides the user with exactly what they want, and more, in a balanced way). Because users have different needs, to be really powerful, a user can tailor their widget settings to their needs (eg: what fields to display, notation to use, etc). Hence, we end up with 3 things: functionality & data sources, user interface, user configuration.
- I mentioned WolframAlpha. Simply integrating it with Kagi Search for widgets might be a good idea. For more specialized widgets (eg: prayer times), it may be easier and much more efficient to simply collaborate with already established services to embed their contents in a Kagi Search results. Basically, why create a calculator widget when an already available service offers that. Just integrate it! (in a widget). This approach shifts focus from developing & designing to picking & integrating. These are two extreme approaches (in-house vs adapted). A middle approach (a mix of those) might be most economical.
Basically by treating widgets as optimized research mini-tools, rather than user conveniences and amenities, can drastically turn a generic search engine into a powerful research assistant.
I dunno. I wanted to suggest the idea and see what everyone thinks.
cheers
