What is a ‘challenging problem’?

Adrian Rezuş
équivalences
Nijmegen
The Netherlands
2016

‘Service to Humanity’
We are, most of us, involved in ‘solving problems’, most of our time. Some
of them are ‘real’ or ‘really important’ (as they say), other ones are merely
‘academic’ (they – the other ones – would say).
A few years ago – sometime in the previous millenium –, one of my ‘polytech-
nical’ friends (an electrical engineer, expert in high power technologies and so
on) managed to circumscribe exactly – to my mind – the first kind: she claimed
she was, simply, ‘in the service of humanity’1 . That’s ‘real’ and even ‘really
important’ (think, e.g., of the poor state of the electrical networks in the States
these days). The other kind is equally easy to exemplify: take, for instance,
the problem Professor Andrew Wiles (well, along with plenty of others, in the
end) was confronted with. since still a teenager: the Fermat Conjecture (now a
‘Theorem’). This one is, definitely ‘real’ – it’s a ‘real problem’, one might say
–, and probably even ‘important’ in some sense; yet significantly less ‘service-
able to humanity’, I suspect. We can certainly survive without Fermat (and,
even, Professor Wiles2 ) for a long while – a millenium or more, from now on.
It is, nonetheless, less obvious that we can survive that long – by our current
standards and habits – without people like my power-planting friend.
One might object to the previous example to the effect that I’m confusing
things like ‘applied’ and ‘pure’ science (maths3 , in particular). Because even
though both my ieee-friend as well as Professor Wiles are actually ‘using’ highly
sophisticated maths, the maths involved are of a different ‘kind’, in each case:
the latter is just ‘pure science’, while the ieee-people are dealing in ‘applied
science’, as they use to call it.
1 This because she was mainly involved in designing / fixing power plants... and / or doing

‘service’ for the ieee-community, in her free time.

2 Sorry, Professor: a sad fact of life, alas!
3 Sic: the plural is British (like in Ancient Greek)! Other people use to think that mathe-

matic (sic) is one!

1
 Wrong idea! How do we know Fermat-Wiles won’t eventually yield plenty
of so-called ‘applications’, within a century or, even, less? Thereby becoming
equally ‘serviceable’ to the humanity next to come (as, e.g., to our grand-sons
and grand-daughters, their sons and daughters, and so on).
Look at the recent whereabouts of the Galois theory: from a purely ‘aca-
demic’ business, this piece of ‘pure’ algebraic wisdom has become nearly un-
avoidable in cryptography, nowadays. And don’t tell me cryptography is no
good ‘for humanity’, or else that it’s only good for spies and secretive persons
or governments. After all, cryptographers – highly theoretical folk, Alan Turing
included – contributed significantly to the end of the Second World War. And
even wars are human affairs, you know? Unfortunately.
“Wait!”, my would-be opponent would likely exclaim. “You’re awfully rhetor-
ical, Sir! This kind of argument destroys already your previous distinguo be-
tween ieee-science – as you’ve put it, in the footsteps of your electrical friend
– versus Fermat-Wiles.”
Not at all! Certainly, my distinguo is ‘timed’ and, above all, ‘empirical’, so-
to-speak. After all, humanity has survived even without ieee-engineers, for a
long while. Past a certain point in time (called also ‘human history’), it won’t be
the case any more, I suspect. But this doesn’t mean the ieee-people are dealing
in ‘applied’ matters alone, while Professor Wiles – and, possibly, a couple of his
friends and fans – are dealing in (mere) ‘pure’ matters. Both ‘kinds’ of science
(ultimately: maths) are based on theoretical endeavours, on ‘contemplation’,
so-to-speak. And, they’re also both based on an ‘appetite of knowledge’, as old
Aristotle would have had it (cf. Metaph., at the very beginning). Usually, I /
we just wonder ‘why’ [such and such is the case]! Sometimes my / our stubborn
questioning – ‘mirare’ in Romanian: it’s easy to find out the Latin therein –
would end up in something ‘serviceable’ for the rest of us, and sometimes not.
Roughly speaking, there is no real point in distinguishing between ‘pure’
and ‘applied’ science (maths, in particular). It’s a matter of time, not one of...
essence.
Ultimately, there is no ‘applied science’ (and no ‘applied mathematics’, for
that matter).
Alternatively, in order to avoid unintended – otherwise boring – professional
conflicts (‘my guild in better than yours!’, and the like), I might be able to refine
the difference suggested above in terms of a single discipline, the calculus, in
maths (also called ‘mathematical analysis’, in old Europe).
Suppose we are confronted – for some reason – with a huge amount of cal-
culations, involving a high maths background. A talented – perhaps even an
average – mathematician (expert in ‘mathematical analysis’, say) would even-
tually solve our problem(s), pending time and / or manpower, by using extant
/ well-known computing methods and recipes. Confronted with the same set
of problems, her neighbour next-door – an expert in tcs (you can even forget
about the t [of ‘theoretical’], here) – would do something else: she will first
re-visit the ‘theory’ thoroughly, rephrase the ‘task’ in her own (guild’s) terms,
extract an algorithm for the task and then ‘implement’ it, as a piece of software.
Ultimately, she’d end up with similar solutions to our original problem(s), ex-

2
cept perhaps for the fact the latter are going to be – awfully – more efficient: in
the latter case it might takes minutes / hours / weeks to get a solution, instead
of days / months / years. Now we are speaking about two ‘applied’ scientists
(and both of them mathematicians, in the end). Yet, if ‘time’ is at premium (as
it is usually, nowadays), only the latter would be really ‘serviceable to humanity’
(at least to a bit of it). For the other one(s) would be always ‘too late’ in this
respect.

Wonder and Challenge

“Perhaps a bit too... Baconian (a reference to our the famous Lord High Chan-
cellor of Great Britain), your power-planting friend!”, my would-be opponent
would, once more, object. “For not every kind of science is ultimately ‘power’,
even if ‘serviceable to humanity’. Even though there is no ‘applied science’,
ultimately, as you timely claimed, you are still confusing the ‘science’ itself with
its specific uses, now and then. As a matter of fact, the recent (2000) Knight
Commander of the Order of the British Empire – as well as plenty of string
theorists, say, to change the running examples for a while – won’t really care
about being or becoming ‘serviceable’, anyway. They’re just... wondering, like
those venerable Greeks you already mentioned would have said. There is always
a kind of challenge, involved. But there is still no ‘service’ intended in actual
problem-solving. And, for sure, no ‘humanity welfare’, or anything like that,
among the reasons a scientist may have to ‘do science’. I’m usually becoming a
‘scientist’ just because I’m curious – or only neugierig, in your German –, there
is an appetite for knowledge, to begin with, indeed.”
Right! This is, is fact, my next point. – I’d rather dissent, however, on
reasons and motivations.
There are at least two distinct ways of ‘wondering about’. ‘I wonder why’,
meaning, more or less, ‘how’ for ‘why’. This is a typical starting point for a
becoming / future scientist. If you ‘wonder’ about the right kind of things, you
may end up eventually with a Nobel prize (or a Fields medal, for that matter),
because you’re ultimately ‘serviceable’ to the rest of us. On the other hand,
you’d never get such a high and explicit (‘human’!) recognition if you’d wonder
about ‘being and beings’, for instance (like Leibniz, long ago, or, more recently,
Martin Heidegger). And only the latter kind of questioners won’t really care
about ‘serviceability’, to humanity and / or whatever else. The former kind
would, at least implicitly.
In other words, while questioning en philosophe, we don’t ‘really’ care about
the world itself, humanity included (you can even write it with a capital h, if
you like: it would amount to the same thing). So the philosopher can – and
will – ask his / her kind of question(s) forever. A (‘real’) scientist won’t even
recognize such ‘questions’ – the ‘philosophical’ ones – as (‘real’) questions.
Because a (‘real’) scientific question is one for which we can – sooner or later
– get (converging, and thus... ‘serviceable’!) answers.
There is no (converging) point in wondering – again and once more, after

3
ages – à la Leibniz: ‘Why is there Being rather than Nothing?’. (Pace Herr
Professor Heidegger, scientists won’t bother about Nothing, anyway.)
This doesn’t mean all ‘philosophical’ (or ‘philosophically looking’) questions
are ‘un-scientific’ (or just ‘pseudo-problems’, as one of our former friends once
claimed). We might usually ask ‘wrong’ – even ‘stupid’ – questions now and
then (‘awfully philosophical’, at a first look). The next generation(s) – one or
more – would ‘repeat’ such questions, in slightly different terms, by refining
them – by ‘getting to the point’, more or less, so that would-be answers would
come, eventually, ‘in sight’.
Whence a kind of (meta-) question I intended to ask from the very beginning
(sorry for the long détour: I was unable to put things in a simpler way):
‘What about the motive ground of a (real ) question? What’s behind a ques-
tion that makes me try, again and again, to get an answer (to it)?’
It’s an offending state of affairs, I’d say. The French would likely call it
a défi, something to be taken, more or less, personally. This comes about to
the English ‘challenge’, more or less. (Britons and most – English speaking –
descendants thereof would rather have it as a kind of club – sport, say – affair.)
Apparently, there is no generic, no ‘metaphysical’ (meta-) answer to this.
Because, simply, why is not every question ‘challenging’ to me? Or else ‘equally
challenging’, indeed? Subjective conditions, knowledge / background, all this is
immaterial, here: I can modify my state of mind, I can also learn, etc. etc. –
Well, if you’d ask me about a specific something, there might be no ‘chal-
lenge’ in your question / problem. Because – simply – I already know the answer:
so – while answering – I’m eventually acting as a mere tutor, as a teacher or so.
No funny business, after all! You can also use a book, instead, with the same
effect, if not even with more profit.
If, however, I don’t know the answer to your question / problem in advance
(and, moreover, if I’m aware of the fact nobody else does), your question /
problem would certainly look ‘challenging’ to me.
Still, there are different kinds of ‘challenge’. Planning to go into this next.
Yet, before addressing the difference, here is a good place – as any other,
perhaps – to add a parenthesis.

A Bad (Meta-) Dream and the Drunk Little Monads

We can first think of ‘science’ as of a vast repository of data, a kind of encyclo-
pedia, let’s say.
If the answer to your question / problem is not out there, in the Universal
Data Base, then your question / problem looks ‘challenging’ to me / to [all of]
us: worth taking the trouble to look into it.
This is a rather superficial view on ‘science’, though.
First, ‘real science’ is highly ‘regionalised’: this would amount to many,
apparently unrelated ‘encyclopedias’ instead.
There are ‘sciences’ not a single ‘science’, indeed: the ‘science itself’ looks
rather like a maze of un-communicating containers, not like a unique Holy Grail.

4
 On the book-metaphor, we are only confronted with disparate chapters in
a novel written by an absent-minded author – at different times –, inspired by
very different, unrelated ‘real-life’ episodes.
On a more ‘philosophical’ metaphor, the true picture of ‘real science’ ressem-
bles to a world à la Leibniz, one without a pre-established harmony, though: a
collection of un-communicating monads, with no Over-All Monad to integrate
– and to harmonise – little poor monads – the folk below – from the above.
The older neo-positivists’ (meta-) dream of a would-be ‘Unity of Science’ is,
at best, a reductive idea, one based – more or less – on the model of a single
God-blessed ‘discipline’: the Theoretical Physics. This was a piece of wishful
(meta-) thinking, in the end. Because we are actually dreamers of many Holy
Grails, as many as the historically attested ‘scientific disciplines’ (if not even
more, since – as we go – we are always tempted to invent new ‘disciplines’ from
scratch, from trifles, or from something like that).
Second, even if granted the (multi-) disciplinary perspective, the would-be –
intended – coherence of a single scientific ‘discipline’ is not always automatically
granted: the ‘real’ little poor monads are often skyzophreniac entities, or else the
monads actually populating our ‘scientific’ world would look like drunk people,
derilium tremendes, rather than lucid, sober, and thus respectable folk. And
this picture reflects – unfortunately – the rule of the game, not mere exceptions.
In spite of all appearences, even the Paradigmatic Queen so beloved by
the earlier neo-positivists – Theoretical Physics – is nothing but a fat, highly
delirious, old lady, nowadays (look at those many, highly imaginative ‘string
theorists’, and to their [highly] ‘theoretical’ zizanies!).
With a popular joke (stolen from the world of mathematics, more or less),
the only ‘unifying principle’ in contemporary physics seems to be the label
‘Theoretical Physics’ displayed on the wall of the physics departments and / or
institutes, ‘round the world. Just like that!4
On the drunkard metaphor, the ‘real’, immediate / urgent challenge would
consist of making people sober, first.
Looking for (a lost) coherence: a very different kind of problem-solving,
indeed! Rather – if compared with usual: punctual / local problem-solving – a
would-be meta-business, so-to-speak.5

‘System’, the Kantian Dream, the Actual Theory-Building,

and the Perpetual Zizanie
The ideal situation for a specific ‘scientific discipline’ is best described by an old
Kantian dream: the Normative Idea.
4 The original maths joke amounts to the observation that the fictitious French mathe-

matician Nicolas Bourbaki has invented a new French singular noun, from a French noun
defective of singular, mathématiques. As a matter of fact, the French plural comes from an
Old Greek [plural] phrase: there was no mathématique in Old Greek, either: ta mathēmatika
were simply ‘the mathematical disciplines’, even for Plato and Aristotle – a venerable example
of skyzophrenia, in this kind of plural(s), we must agree.
5 This has nothing to do with ‘metaphysics’, ‘meta-physics’, and the like, by the way.

5
 In ‘real life’, this points out to something like the ‘systematic character’ of a
(particular piece of) science: in a given ‘discipline’, things must ‘hang together’
first: there is no ‘science’ otherwise.
Whence, the meta-business alluded to above, a normative idea (‘look for
coherence first!’) should come before the actual (punctual / local) business of
problem-solving. If our ‘science’ (‘scientific discipline’) consists a mere bric-à-
brac collection of conflicting data and statements, there is little chance we can
ever solve local / punctual problems in this area; the said ‘science’ is, in fact,
useless.
Aside. There is a famous general theorem on computation, in Lambda-
Calculus (due to the father of the business, Alonzo Church, and to one of his
students, J. Barkley Rosser), the so-called Confluence Theorem (also known
as the Church-Rosser Theorem; CR, for short). Roughly speaking, the CR
Theorem says that the result of any particular computation [3+(2×3)+(5×2),
say], in a reasonably ‘coherent’ computation system, must not depend of the
order of performing smaller computations steps. So, it would be reasonable
to ask from a specific computation system to satisfy this kind of (minimal)
requirement [here: confluence, in technical terms]. Indeed, imagine the disaster,
otherwise: a kind of algebra / arithmetic where 3+2+1 would lead to different
results, according to the path of computing you might choose: ‘do first 3+2’ or
‘do first 2+1’. – This is mainly to stress the fact that a ‘coherence’ requirement
might not be a mere [‘traditional’] logic requirement, where coherence is taken
to mean logical non-contradiction or consistency.
The meta-business above – characterized by the (more or less Kantian ‘nor-
mative’) recommendation / requirement: ‘look for coherence first!’ – has been
always among the main concerns of ‘ordinary’ scientists; so obvious that it was
not even worth mentioning as a separate ‘scientific’ task. Whatever the philo-
sophical ancestry (Aristotelian, Galilean, Cartesian or so), it is only in recent
times that ‘professional’ philosophers managed to ‘take over’, and grab the sub-
ject for their own. The modern catchword for this kind of [meta-] concern is
‘epistemology’.6
In ordinary scientific practice, the ‘normative’ requirement, mentioned above,
would amount to the fact that any ‘scientific discipline’ should have a ‘system-
atic character’; specifically, to the fact that any ‘particular science’ should be
presented / organized as a ‘deductive theory’.
Initially, the implied – more or less managerial – task looked like a paradig-
matic business for (‘professional’) logicians. In a longer run, in spite of an
endemic – oft much too loud – agitation (around the early thirties till the late
forties), in both Europe and overseas, it quickly turned out the logicians ‘proper’
have actually little – or nothing – to say about the subject itself. In retrospect,
the extant contributors to the so-called ‘logic of science’ have done more damage
than otherwise.7
6 In plain English: ‘theory of science’. To be fair, there is no ‘theory of science’ so far, there

are just some – otherwise rather smart – people around, thinking ‘theoretically’ about such
things and trifles.
7 Honestly speaking, ‘logicians’ are not and should not be concerned with such trifles! ‘From

6
 In the meantime (during the last half a century, say), we managed to learn
/ acknowledge a few more things, nevertheless.
If the main (meta-) business in ‘science’ – specific ‘disciplines’, I mean – is
the so-called Theory-Building, then we must take for granted the actual state
of affairs (and, possibly, remember that this was always the ‘current’ state of
affairs in ‘science’, along the ages, at any specific point in time), namely, the
fact that there is no ‘normal science’ (to use a recent catchword), except for
a while or two (!): the (historical) rule of the game is, rather, the ‘theoretical
conflict’: a nasty thing to cope with, indeed!
In other words: ‘true / real science’ is, in fact, a Perpetual Quarrel among
Dis-Agreeing Factions – it’s actual locus is a Battle-Field, down-here, rather
than a Serene Privileged Room, high-there, in the Ivory Tour.

Epistemic vs Technical Challenge

Coming back to ‘challenge(s)’, my concern in the title. –
There are plenty of things to say about.
Let me first display a few examples of ‘challenging’ questions / problems in
my sense (in logic, my preferred spare-time, for a rather long while).
Rather elementary, I’d say (the required background can be reviewed / ex-
plained in a few minutes – and /or a few lines in print).
[A conceptual ] Aside. If I remember well, Henk Barendregt pointed out,
in conversation, already thirty years ago or so, to the fact that my colloquial
uses of the (epistemic) qualifier ‘elementary’ (as applied to maths problems,
mainly) might [have] look[ed] oft [those times] rather loose and even confused.
The point concerned standard – yet careless – ways of of speaking in maths: PA
(Peano Arithmetic) was / is still usually referred to as ‘elementary arithmetic’.
A very wrong idea, if you really care about ‘complexity’ matters and things
like that! So, I’d rather think twice or more – since –, before referring to a
specific PA-question as being ‘elementary’. A bit later, I had the opportunity
to dissert at length about ‘colloquial’ proof-qualifiers like ‘obvious’, ‘easy’, ‘as
ever’ etc. [in the introduction to Abstract automath, 1983, kind of a digest,
recording endless conversations with Nicolas G. de Bruijn, in Eindhoven, on
proving in mathematics and so on], and even managed, next, to put together a
little private theory of the would-be ‘degrees of elementarity’ in mathematical
practice. (Whence also my disappointment as regards the current talk about
‘complexity’ in logic, maths, tcs, and the like.) – Here, I’d rather avoid such
subtleties, however. In particular, I’m going to use the qualifier ‘elementary’
only as regards the way of formulating / expressing a particular problem /
a logical point of view’, if ‘physics’ – or ‘mathematics’, for that matter: set-theoretically re-
shaped as ZFC, say – would turn out to be ‘inconsistent’, then much worse for the ‘physics’ –
and / or the ZFC-based ‘mathematics’! Why bother? For – remember, please! – these are...
worldly affairs logicians are not and should not be concerned with. – The fact is that recent,
serious logicians would typically avoid the subject, nowadays, half a century – or so – later,
the associated message being, roughly said: ‘mind your own business, folks! we don’t really
care about yours!’, etc.

7
statement, not as a qualifier of the would-be methods / techniques etc. that
might be needed in order to solve a problem and / or to prove a particular
statement.
[1: The Belnap story] Suppose you would have asked me, some thirty five
or forty-so years ago (’round 1975 thus), to prove the following little puzzle (in
‘propositional logic’):

Let [B], [CB], [I] be formulas as below, and BB’ be a ‘proposi-

tional logic’ with implication [→] only, axiomatized by [B], [CB]
and modus ponens for → (substitution rule tacitly assumed).

[ B ] (p→q) → ((r→p) → (r→q)) – [prefixing]

[ CB ] (p→q) → ((q→r) → (p→r)) – [suffixing]
[ I ] p→p – [identity]
[ mp ] A→B, A ⇒ B – modus ponens, detachment for →.

Show that no substitution instance of identity [I] is derivable in BB’

(idest: there is no proof of [I] from the [B] + [CB]-axioms, with
modus ponens and substitution). –

This is a rather ‘elementary’ problem in (propositional) logic. Easy to state

for everybody, no ‘advanced’ knowledge required to grasp its meaning, etc.
Yet, ‘challenging’, indeed! Thirty five or forty years ago, I won’t have had
the slightest idea on how to prove the claim.8
I’d classify this querry as an ‘epistemic challenge’ (it is, more or less, clear
why).
The specific point, here, is in the fact that the problem can be solved in
different ways, and at different ‘levels of abstraction’, so-to-speak.
This being said, I won’t mention the [associated] bibliography, anyway.9
[2: The Tarski story] Suppose somebody [else] would have asked you some
85 – resp. 91 – years ago (i.e., in 1925) to prove the following equally puzzling
(meta-) statement:

Let [K], [D] be (purely implicational) formulas, in a propositional

language as above (→ stands for implication).

[K] p→(q→p) – irrelevance

[D] p→(q→((p→(q→r))→r)) – pairing
8 Historically, ‘the Belnap conjecture’ – now a ‘theorem’, due to Nuel Belnap Jr – stated

above remained, actually, a very stubborn open problem for more than fifteen years. At the
time of writing there are three distinct ways of showing this. In each case, the techniques
therein involved are surprisingly ‘advanced’, in maths terms. No connection whatsoever with
the popular wisdom on ‘propositional logics’ you may eventually learn about, in a first year
logic-course, in maths and / or philosophy departments, etc.
9 Just try the problem ‘by hand’, without cheating (looking on the web etc.): if you end

up with an answer in less than a year, say (!), then you are a smart mathematician, likely
(although – to be fair – you won’t thereby contribute a bit to the welfare of humanity, alas).

8
 Let L be a finitely-axiomatizable propositional logic (i.e., a logic
with finitely many axioms) and modus ponens for → (substitution
tacitly used, as ever). Show that L is axiomatizable with a single
axiom (modus ponens, and substitution), if both [K] and [D] are
derivable in L.
Historically, this statement was claimed in 1925, without proof, by Alfred
Tarski. The Pole is, among other things, the founder of modern Model Theory, a
branch of [contemporary] Mathematical Logic, by the way. Parenthetically, the
first proof of Tarski’s claim was obtained by the undersigned, in 1979 [published
1982], using a technique involving concepts not available in 1925 [λ-calculus:
invented around 1931–1932, in the United States, by Alonzo Church].
Aside. A meta-puzzle: how did Alfred Tarski manage to get such a result,
in 1925? The meta-puzzle points out to a famous, and very similar, in fact,
‘epistemic’ situation, namely, to the case of Fermat’s claim. If Fermat did really
have a proof of the ‘Fermat theorem’, his proof was, definitely, not like the one
Andrew Wiles was able to produce a few centuries later. Wiles’ proof won’t fit
the ‘margin’ of that famous Greek book, either; but the real point is in the fact
that it won’t fit the overall epistemic context, anyway. Because Professor Wiles
was inheriting of a very different kind of mathematical wisdom. Pierre Fermat
couldn’t have ever dreamed about elliptic curves! At least not in this context.
In the end, if we are to be ‘really fair’, Andrew Wiles has not actually solved
the original – ‘historical’ – problem; namely: ‘prove the Fermat Theorem with
the conceptual means of his time’! –
On the other hand, my own ‘meta-conjecture’ would consist of saying that
Tarski anticipated – somehow – ‘λ-calculus’, ‘combinatory logic’ and the like.
Because there is no other – reasonable – explanation: we cannot prove Tarski’s
claim by essentially ‘other’ means!10
This is also an ‘epistemic challenge’, like the one under [1] above.
Specific point, here: very likely, this problem can be solved in only one way
(mine!) and Alfred Tarski was apparently aware of it, avant la lettre. (How?)
No bibliography, again.11
[3: The Wos story] Suppose now you are a member of the relatively new
[scientific] AAR-community (the Association of Automated Reasoning) – so you
are already familiar with various AR (Automated Reasoning) software packages
(otter, etc.) and / or techniques of [soft-] proof –, and the current President
of the AAR (actually, it’s Larry Wos, these days12 ) would have displayed –
at a recent AAR-meeting – a rather long formula – [R] say – containing circa
hundred symbols or so, claiming [R] is axiomatizing classical propositional logic
CL, with modus ponens (for ‘material’ implication) alone. The Wos Question
/ Problem is: show this using AR software (idest: prove [R] from a particular
axiom system for CL, usig otter, for instance).
10 Well, all this would leave Fermat as mysterious and as cryptic as he has been so far, I’m
afraid.
11 If you manage to do it ‘by hand’, without ever mentioning λ-calculus, combinators and

the like, then you’re certainly at least as smart as Alfred Tarski!

12 These notes were written in September-October 2010.

9
 Rather tricky, I would say, if you are not already familiar with the Tarski-
story, already mentioned above, under [2].
Here, formula [R] (they call it a ‘Rezuş formula’ nowadays, although it
should have been ‘Tarski formula’, as a matter of fact) can remain anonymous
(you can easily obtain explicit examples thereof from the solution to the previous
problem, anyway).
The ‘Wos Question / Problem’ is, however, not of the same type as under
[1] and [2].
This question / problem is, rather, an example of a ‘technical challenge’.
Because associated with a specific ‘technical’ constraint (here: ‘use AR soft-
ware’).13
Acknowledgements. I owe the last distinguo above to Branden Fitelson (now
[2016], at Northeastern University, Boston, MA), who pointed out the fact that
(generalizing a bit), actual conditions / constraints occurring in our current
problem-solving endeavours may not be always ‘purely epistemic’, but rather
– and typically so, in ‘real life’ – of the kind claimed by my power-planting
friend, the ieee-engineer, a while ago. Moreover, such (‘real-life’) constraints
would usually make things less obvious than they might appear at a first look,
in abstracto (‘laboratory conditions’ and so on). — The latter distinguo –
epistemic vs technical challenge – deserves further reflection. Putting things in
slightly different terms: ‘does common science need... translation in order to
become serviceable?’ For, if so, this would involve a different kind of knowledge,
apparently!
Thanks are also due to John Halleck (University of Utah, Salt Lake City,
UT) for making comments, similar to those of Branden Fitelson – including a
huge amount of technicalities on so-called ‘Rezuş formulas’ and the like.

“I think what we have here is a difference of cultures, and cultural

context. The context in which Dr. Wos is publishing is people
working on automated therorem proving programs, trying to make
them solve a wider class of theorems. Dr. Wos is himself famous
for ‘pushing the envelope’ and developing techniques to do just that.
[...] At the current ‘state of the art’, there are huge areas where there
are human proofs, and automated theorem proving does very badly
and either can not solve them at all, or only some programs can.
His [Wos’] challenges are not intended to be particularly challeng-
ing to people, and certainly not to any half-way educated person
with access to the literature. What they are intended to be is
something that is very difficult for programs to do, and are areas
where research in methods is sorely needed. And the Rezuş axiom
for BCIW, while never intended for human display, is perfect as a
test of an area that (as he pointed out) computer programs are very
weak.” [JH 20100811]
13 Actually, at the meeting of the AAR, in June 2010, Larry Wos gave a different, more

complex, example, based on the same Tarski-story as above. I confess I didn’t first see the
‘challenge’, whereupon I certainly owe him an apology!

10
Nicely said! Although I’d have rather mentioned a ‘diference of mathematical
culture(s)’, instead!
Postscript [20101003]. Apparently based on a previous approach due to
Dolph Ulrich (2004–2005), John Halleck has provided, a few days ago [20100923],
a genuine solution to the `challenging puzzles’ [2] and [3] above. Actually, Hal-
leck did not prove [2] ‘by hand’, the old-fashioned way; he used specific software
instead, enabling him to produce derivations very close to the Lukasiewicz style
derivations (modus ponens cum substitution) of the late twenties and early thir-
ties. For a short λ-calculus version of Halleck’s proof, see the Addendum to my
note Tarski’s Claim: Thirty Years Later [October 1, 2010].
Nijmegen: September–October 2010, rev. May 12, 2016.

c 2010–2016 Adrian Rezuş (Nijmegen, The Netherlands)

c 2010–2016 équivalences (Nijmegen, The Netherlands) [(pdf)LATEX]

typeset by romanianTEX c 1994–2001 Adrian Rezuş

Draft: September 9–11, 2010
Revised: October 3, 2010; April 5 2016
Last revised: printed in the netherlands – May 12, 2016

12