These robots weren't really "walking" in the sense that humans walk through continuous dynamic balancing, i.e falling forward.

These used quasi-static walking, where the zero moment point (like a moving centre of gravity) is kept within the support polygon of the footprint. This is what gives them their weird swaying gait and extremely conservative movement characteristics. You could never make a bipedal robot run, jump or respond to large and sudden external forces using this method. It's essentially a balance free movement hack.

Computer science research tends to look like this. Take a seemingly ambitious idea, then spend eons making a version of that idea which doesn't scale and probably doesn't work. But no one is really sure how far this incomplete idea will go. After demonstrating it we realize what the limits and next steps are.

MIT's Leg Lab was doing dynamic walking around the same time. http://www.ai.mit.edu/projects/leglab/robots/robots.html

Ya, they walk like old people. By keeping thier cog over thier feet they are able to stop at any moment without tipping over. That's how old people with diminished motor neuron function walk. Both play it safe because they know they lack the reaction time to prevent a fall once cog is outside their footprint. It is also how one walks when on very slippery surfaces.

According to the article E0 was static, E3 was dynamic.

What none of them did, however, was “learn” (as the title suggests). They used hardcoded algorithms.