Jesperson's dictum needs an additional criterion.
It should be the easiest to learn that also can express what people might want to say with it.  I mean, here's a language that has only one word "Hodor".  It means "whatever".  It's easy to learn, but useless. Ease of expression is folded into that criterion--if it's too hard to say, you might not want to say it. An oligosynthetic language could be easy to learn and capable of expressing a broad range of information, because it builds all words out of a few simple elements, but if it is poorly crafted it might be so hard to use it to say anything that people would rather draw pictures and use pantomime. 
Jesperson is right to an extent.  The usefulness of a language comes from other people knowing it and being able to use it to exchange information with them.  Simplicity lends to usefulness to the extent that it increases the number of people you can exchange information with, but that must be multiplied by the value of the information it can be used to exchange.  Just being able to say "Hodor" won't do any good, even if everybody can do it.
Having a base of speakers gives a language a leg up on usefulness, regardless of its own features.  That's why Ido didn't supercede Esperanto which is itself eclipsed by English.  It's like lighting a fire.  I think if a simple enough language becomes also widespread enough then it will start to snowball, the fire will be widespread and hot enough that it is exposing lots of fuel to ignition temperatures, thus it will also have a high spread rate.  And relative simplicity should itself be a factor in early going because it helps persuade people to try it out. 
Also finiteness.  I think this is a vitally important selling point.  It must be possible to learn the whole language.  There should be an end state where nobody who has studied the language longer, or been more engaged with developments, will be able to suprirse you with stuff you don't know.  And that's why the language should be oligosynethicish, and the set of allowed sounds should equal the set of root words.  Which is why it must be largely apriori, though it can follow natural language where it can (why not?). 
That way a person who has just learned all the basic elements will know all the words that might be used by a lifelong speaker, kinda theoretically.  One of the things that went wrong with Esperanto was that it kept growing, adding new words like a natural language.  I believe there are now often multiple words for the same thing, and when you encounter them you won't know what's being said unless you have learned that version of the word.  This is a feature of natural languages, especially English.  Oligosynthetic languages would have it too, many different ways to combine the basic vocabulary to express the same idea, but all such expressions would be made of the same elements.
This importance of finiteness is why an auxlang based on an existing natural language shouldn't be too close to the natural language itself.  Because then it's just a limited vocabulary version of the natural language, and you'll get words outside the restricted vocabulary sprung on you because people think you know, say, English, because you speak Basic English or German because you took two years of it and can read signs and order dinner.  Being oligosynthetic is itself enough to make an artificial language radically different, while also improving finiteness, though it adds some to complexity (but on a scale that pales in comparison with the learning of vocabulary).        
On 03/18/17, Leo Moser<[log in to unmask]> wrote:

Two statements on simplicity and IALs:

1. "The best international language is that which is easiest for the greatest number of people." Otto Jespersen.

2. “English is far too difficult to be an appropriate IAL.”;


I strongly disagree with Jespersen’s dictum, but fully agree with the second statement.

The value of a language is based on the uses to which it can be put. My dictum would be:

"The best IAL is that which is most useful for the greatest number of people, worldwide."


People learn languages because they are useful, not just because of their simplicity.

As difficult as it was, Volapuk showed that such an IAL could be somewhat useable.

People who adhere to Esperanto tend to find its circle of samideanos to be useful.

They put up with its long identified difficulties because of such things.


Ido was designed to be easier, and was. But it did not supersede its forerunner.

IALs like Frater, Glosa, etc. were perhaps simpler yet—Toki Pona is far simpler.

But the push for greater simplicity can disrupt usability in certain domains.

The global language should be the most useful for all persons and purposes.

This means world literature, commerce, science, law and diplomacy, the internet, etc.


Right now, English seems to be fulfilling that purpose . . . to a degree.

But it remains needlessly difficult, takes almost a lifetime to learn well.

English is also tending to split into regional variants.

(Consider its complex and varied vowel patterns!)


English is IMO inappropriate as the long-term IAL.


I have great respect for Jespersen, but I believe his famous dictum must be discarded.

An IAL can be complex in certain ways, if that complexity adds enough to its usability.

Much of the complexity of English detracts from its usability.


Best regards                LEO