Netscape 1.x

The model of licensing software was pretty typical for the time. I disagree that they didn’t have a sustainable business model. Sure, IE4 was better and Netscape didn’t respond very well, but by 1997 it’s market share and revenues were already significantly eroded.

Mozilla was based on gecko and related technologies which were being developed for communicator, but when the revenues associated with the browser evaporated, there was no point in investing in it any further. Rather than simply letting it rot we decided that the way to stick a thumb in Microsoft’s eye was to take one step further and not just make it free, but make it open, with an organization to perpetuate open browser technologies. But everything that went into Mozilla was already long in development.

That said, IE 4 was a better product - which goes to show an unlimited war chest, a singular drive, and no commercial necessities can go a long way in developing software. They also had a lot of layout experience internally with office, and made smart key decisions that we had blundered on early due to lack of layout experience. We were mostly an internet protocol company, and our layout experience was fairly simple. That translated into a really key early mistake - when calculating table layout, which were the primary way of organizing visual elements at that time, we assumed tables were infinitely large and as we loaded elements and calculated their size we sized the table to fit them. Microsoft did the opposite and assumed they were zero sized and rendered each element in lay out as its size is known. This allowed a perception of faster loading because you could see the loading as it progressed. Unfortunately the layout engine in early Netscape browsers was so deeply enmeshed in the code that removing it and replacing it with gecko was a multi year effort, and we were being driven my marketing and product to release versions so as to not be seen as falling behind.

Ultimately that was the downfall of Netscape - product had way too much power over engineering, and while we had the best engineering team I’ve ever worked with in my career (including at multiple FAANG, unicorns, Wall Street quant groups, etc), we were constantly hamstrung by product chasing some short term priorities to keep notching differentiating features. Microsoft had a very clear goal - eliminate Netscape’s browser revenue at any cost - and the ability to redirect most of their engineering talent into that goal. Once they released a bundled browser that was technically superior as a web rendering tool for absolutely free (not just free, but mandatory and inseparable from their OS), it was all over for us. All the differentiating features were buggy baggage on a partially completed product vision that had no revenue model - so, the product never finished.

The key is there -was- a business model, and it was successful. But no one can compete against a monopoly price dumping to zero. Given another six months we would have had the new layout and rendering engine in place, and would have had the better browser again. But with no revenue possible, the only option was to fold. But the smarter option was the one we took, and thus set up the precedent for large open source transfers and foundations. And, IE languished and rotted, because why bother investing in something that makes no revenues?

If your customer was cheap and needed more bandwidth you could use EQL[0] to load-balance a PPP connection over multiple POTS/ISDN lines if your ISP supported multiple connections. I set this up for a number of businesses back then when office staff started arguing with each over who was using all the bandwidth from a single 28K modem.

[0]https://www.kernel.org/doc/Documentation/networking/eql.txt