Re:Is the web getting slower?
by WrtCdEvrydy on Wednesday September 09, 2020 @12:01AM
(#24416965)
To be honest, we have decided that lowering the cost of building software is the best priority so we choose more and more abstracted software that requires more resources.
The javascript cascade from hell starts when you add a single ad. We wrote a test param to remove ads and the page load went from 35Mb to 1.2Mb... all of it was JS and ad tracking code.
34m ads on each page load? Why not use an ad serving network?
Re:Is the web getting slower?
by WrtCdEvrydy on Wednesday September 09, 2020 @08:30AM
(#24420258)
Yes. Money and Greed.
Edit: Server-side and client-side rebidding.
Product managers add features and/or speed up development by using more resource-hungry runtimes up to the point where the users still tolerate the performance but barely so. Once the product has enough traction and switching costs, the user would rather tolerate the mild pain of slowness than the more acute pain of switching and adapting to a (so far) faster competitor.
This is very much like the idea of the optimal price point where the customer already cringes but still buys.
The behaviour of desktop software developers in 1990's was a prelude to web frontend and app developers today. As the resource capacity of PC's increased through the 90's, software developers grabbed it for themselves; for the end user, the experience of using the software never got any faster.
As the storage, processing power and network speeds have increased, web frontend and app developers seem to have grabbed it for themselves as well; the end user experience has generally not gotten faster (if the user follows the instructions: use popular, Javascript-enabled browser, enable images, video).
In the 1990's, users were asked to upgrade their computers so they could run the new, larger but not experientally faster, desktop software. Today, we see the same sort of OS and device upgrades; "unsupported" versions of software that will not run on newer version of the OS. Plus we have a culture of allowing automatic, remotely-controlled installation of new software. Automatic upgrades. Telemetry. "Field testing" of new features. A sizeable amount of user choice and necessity for user consent has been successfully eliminated.
The number of outgoing TCP connections made by today's average websites, usually triggered through cascades of Javascript files, is simply staggering. It would never have been acceptable to do that with the networks and server software of the 1990's and 2000's. Users were not involved in the decision of developers to grab the newly available network resources and CPU for themselves.
I do not use a popular browser for most web use. This allows me to only make a single TCP connection to a website when I access it and can retrieve dozens or hundreds of pages over a single connection, thanks to old, standard features of the web documented in RFCs (pipelining). It is very fast and efficient. For me, the web has gotten much faster, but only because I choose the software and I decide whether and when to upgrade. It is my computer, after all. I am not really impressed by the popular browsers of today because it does not feel at all like the user is in control. Users of these programs have been reduced to guinea pigs.