>I dislike the fact that Clojure can, in a sense, be written forwards or backwards.

If you're referring to the distinction between -> and ->>, it's not about direction preference, but rather that functions that conceptually work on a single subject (e.g., conj, assoc) typically take that subject as its first arg, whereas functions that work on a sequence subject (e.g., map, filter) typically take that subject as its last arg. From that consistency emerges the two different threading macros.

The parent is almost certainly referring to the distinction between:

(-> (initial) a b c d e)

and

(e (d (c (b (a (initial))))))

I find -> and ->> useful, but the new threading operators kind of silly; they're very easy to implement in terms of -> and ->>, or in terms of higher-order functions, and don't really cover all the use cases one would want---it offends my sense of aesthetics to have them (in their particular implementation) added to the core ns.

(-> (initial) a b c d e) is like "The city on the bay in California"

and

(e (d (c (b (a (initial)))))) is like "The Californian bay's city".

These variations are different thematic structures to language.

In my view it's exactly the opposite:

    (-> (california) bay-in city-on) ; "The Californian bay's city"

vs.

    (city-on (bay-in (california))) ; "The city on the bay in California"

The former ("->") starts with what's fed into the pipeline, the latter starts with what you get out of it. In addition, "->" avoids the nesting of expressions regardless of the number of steps, which might be useful in some cases. But both styles have their place.

Use it every day and you'll probably think different about it.

(baz 33 (bar 12 (foo 33 (+ 1 42))))

Is way harder to read than:

   (->> (+ 1 42)
        (foo 33)
        (bar 12)
        (baz 33))

I totally agree and personally love using -> and ->> but I've heard people say that the idiomatic way is to use let instead:

    (let [fortythree (+ 1 42)
          foo33      (foo 33 fortythree)
          bar12      (bar 12 foo33)
          baz33      (baz 33 bar12)]
       baz33)

I understand the logic behind preferring let and a lot of times it really does make things clearer to have everything named and explicit, but especially in prototype code, I find the need to name things to get in the way. I would prefer not naming things over badly named things, for example, and that's usually how I decide between let and thrush (all things being otherwise equal).

There is a lot of stuff in Clojure that can be done in more than one way. I know that Python focuses on "only one way to do it", but Clojure's focus is on simplicity, as defined by Rich Hickey in Simple Made Easy [1], and that is a much better focus. I find that, as Rich says (and I'm paraphrasing), simplicity is by far the most important concern when programming. It dominates all other concerns in terms of importance and payoff. I believe I have become a much better programmer since I learned Clojure and listened to Rich's talks.

[1] http://www.infoq.com/presentations/Simple-Made-Easy

> Clojure can, in a sense, be written forwards or backwards

Clojure can in many senses be written forwards or backwards. By defining this short macro somewhere...

    (defmacro topdown-tale [& stmts]
      `(do ~@(reverse stmts)))

we can write our top-level control statement at the top and the functions it calls underneath, by bracketing with...

    (topdown-tale
    ...
    )

Many languages are subject to such assymetries: at times one writes nested calls, and at others, chained calls.

    a(b(c()))
    a().b().c()

the latter form requires less mental overhead so why not favour it whenever possible?

It's more like c().b().a() which shows there is some "overhead" right there.

I dont think so... c() would be evaluated first which is inconsistent with the other examples

Edit: I realize vemv's examples are inconsistent as well