I've been thinking very similarly for a while, and especially after YouTube's tie-ins with previous debates that didn't use the medium very appropriately.

The traditional debates are a kabuki-choreographed mess, with their negotiated formats and stuffy moderators and artificial limits, preventing almost anything of substance from being discussed.

I do think a firm set of time limits, and expected presentation format of replies-chained-to-each-other, would be important to make the process understandable to candidates and voters – especially at first. Viewers and candidates need to know the boundaries of the engagement, even if they are free to offer unbounded supporting material elsewhere.

A keep-it-simple first MVP could be roughly the "Lincoln-Douglas" one-on-one debate format, as practiced in competitive high-school/college debates. That is: a series of fixed, balanced time slots alternating between the candidates, totaling 35-45 minutes. (Direct live cross-examination would be out, given the intentional production lags, but knowing the presentation-order to voters, candidates could challenge opponents with pointed questions, then follow-up in their next video about how the question was dodged. Supporting video/audio clips and graphics/animation would be welcome.) In this form, I believe it'd be best to only release any when the full set of videos is available. Then, they can be binge-watched as a cohesive discussion - rather than picked-apart and micro-adjusted with regard to the daily campaign news scraps.

Of course that only works for 2 candidates – but when there's more, maybe round-robin?

After some comfort with the general idea, other less-linear 'hypertextual' presentations like you suggest could be introduced. Still, the value of a 'debate' is the shared-context-narrative of challenge-response, multiple levels deep. Adding too many forking paths or remix options would dilute that.

Another possibility would be to give candidates a total time-budget as with a chess clock, leaving them free to set their own sizing/pace of individual exchanges.

Or run it as a lengthy 'tennis-volley', each day with one new video from alternating campaigns, each expected to respond to the preceding series and comment on new developments.

This online canned debate concept appears beneficial at first glance but is missing some desirable components. It would be good to know who is involved in writing the texts to be recorded by the candidates and perhaps even more desirable to know which existing clips have been seen by the actual candidates prior to preparation of responses.

Existing live debates demonstrate that candidates themselves are hearing the opponents' arguments although they do not require candidates to write the content of rebuttals or reveal who has written them. But unusual challenges can still require immediate and unprepared responses. That dynamic would sadly be lost in this design and thus it is a dream for campaign strategists.

Of course strategists who negotiate rules for live debates do their best to avoid the need for any last minute thinking by candidates. IMHO an improvement would be to counteract these strategists with rules and some of this is happening.

Treating elections as a game entirely to be played by strategists is to me a dark and sinister vision ( and a mostly applicable one ) that leads to responsible complaints that both parties are acting alike and individual voters decisions have become insignificant.

As Brad points out, this sorry impression matches the understanding of the strategists who will focus on very small populations. Strategists including Brad have recognized that relatively few persons' opinions actually matter.

Anyone can see these categories of voters who matter are subject to efficient focused manipulation. Meanwhile all voters are also subject to manipulation by strategists who explain that most of their opinions do not matter and even ought not matter when the incumbent is apparently extra awful.

That perspective is one that leads would-be voters, regardless of intelligence or good intentions to stop caring about and even avoid voting - and that matches my overall personal reaction to several of Brad's fix-the-vote scenarios.

I would rather fix the system that leads to these highly optimized non-democratic strategies.

Perhaps some Trump voters were hoping to "drain the swamp" of this kind of campaign strategy. That would have been my single reason to vote for him. But there were so many reasons not to, many of which were revealed only after the election.

My favorite improve-the-election rule is to require publication of short lists for cabinet positions in campaign materials.

My next favorite is to allow voting for more than one candidate (approval voting) particularly in primaries.