Comment by wolpoli
2 months ago
> A lot of classic software essentially worked more like a database. In the last 10-15 years it's all moving to an algorithm.
You just described what I missed about the older software. Older software gives users control over sorting and show data in a tabular format. Modern software sorts data with an algorithm, with ads mixed in, and shows data in a card format, making it a lot less usable.
Exactly. My related observation: half of the SaaS products I see would be more useful and ergonomic for the user if they were implemented as an Excel sheet.
(I actually worked for one of such "better off as an .xls file" startup in the past, and its main competitor was an incumbent that sold the same stuff as an Excel extension. Trying to replace that with a React app is not a worthwhile use of life.)
Algorithms are fine. I'll happily apply the most advanced ones I can get. The problem is with who applies them to what - as you and GP said, it's about user control - or, currently, lack of.
Excel is great until it you need to do something that takes up more space than a single screen. Then it isn't.
Sending sqlite databases to the users which they can interact with using both sql and a viewer is where it's at.
I wonder if being a more powerful backing data store for Excel is one of the remaining reasonable uses for Microsoft Access, at least for users of the Windows desktop version of Office? Access is still included in many editions of that (although not on Mac or web or mobile). Officially it’s even still supported and not deprecated, although of course it’s very much not emphasized by MS any more.
Other options might be SQL Server Express or SQL Server Express LocalDB, the latter of which seems conceptually very much like SQLite within the MS ecosystem, and both of which are usable for production purposes at no cost within the technical limitations that differentiate them from paid editions.
The SaaS I am now working for is a react app + some fancy intelligence.
We are not front-end people, so the app is built with the expectation that people will be doing their filtering, searching and using the intelligence we provide, but in their home turf (excel).
Our app also lets users "track" certain events, and we do not use push notifications, rather we respect our user's attention and email them a short summary, and link to a csv that they can use!
> Older software gives users control over sorting and show data in a tabular format
I'm old and tend to agree, but I suspect this is similar to "you used to have a knob on the TV that showed the channel it's on".
That immediately made me think of the digital TV switchover. The elderly father of a friend of mine would spend much of his time in front of the TV, and could operate it without assistance, thanks to the simple 1:1 mapping between buttons and functions.
After the digital switchover, there was now a set-top box, and electronic program guide and three-figure channel numbers thrown into the mix, as well as stateful aspect such as whether the TV was set to AV or still trying to use its now-obsolete tuner.
For someone with poor eyesight, limited feeling in his fingers and limited ability (and admittedly willingness, too) to build a mental model of how the menus worked and how they can be navigated, it spelled the end of his unsupervised access to TV.
The big difference for me between database-query-driven and algorithmically-driven is that the latter makes it very hard to know when you've completed an exhaustive search. Indeed for the likes of meta and tiktok that's a feature, not a bug, since their goal is to keep you engaged and plugged into "their" content forever.
I have a "Roku TV". It has many pointless problems. But the biggest is that it takes a very long time to turn on. If you turn it on by remote control, there is no indication that anything has happened until the TV finally starts showing something. You just have to hope that the signal got through. It's fairly frequent that we fail to turn the TV on because we assume that it just isn't done getting ready to turn on yet.
You can avoid this by using the physical power button, which is conveniently located behind the TV. It will still take forever to turn on, but there's no ambiguity over whether you've started the process.
I still have trouble believing the device was allowed to reach customers in this state.
Not every change is for the better. You gotta admit that TVs used to be able to switch channels much faster then they do now. And analogue controls in cars are safer and better then touch screens for everything.
A lot of change is for the better, but quite a lot is a regression.
"Better" and "regression" are purely subjective.
I have access to around 1000 "channels", if you include live broadcasts and network-like apps. How exactly were old TVs better at helping navigate that?
5 replies →
The problem is with cataloging and discoverability. After a certain number of photos (or movies, songs, whatever), finding what's relevant is non-trivial.
Or the problem is what's relevant, such as Spotify's 'what the user likes and is in our bargain bin for plays'. If the photo algo wants you to get multiple prints which mostly happen when you are sending them to others it's going to push group photos and baby pictures not that epic moonrise.