From time to time appears in typology forums some kind of debate about which function is "the best". This is a pointless debate. Best implies a subjective valoration, there's no objective answer. The most we can do is discussing which funtion is more useful for achieving a particular goal, or developing a particular task, according to a previously defined measurement criterion (time spend, quality of the final product, whatever). Even in these situations, a function alone could achieve nothing. I will exemplify this with Ti and Te.
But what is Ti, and what is Te? No debate about this is intended (here) so I will use the simplest definition that could be valid under most interpretations of the problem, and at the same time could be quite well related to reality (external phenomena): Ti = objective static conceptualizator; Te = objective dynamic conceptualizator. With these definitions in mind, we can easily see the correlation about Ti and logic, Te and empirical data. These are not synonyms, but the latter are consecuences of the nature of the functions. Ti is not logic, but it's the best equipped for developing this kind of task. Premises are well defined and inmutable concepts, so are conclusions, and logic stablishes well defined relationships between them...
So when any of us tries to solve a problem, which method would be superior, Ti ("logic") or Te ("empirical data")?. Void question, we need of both of them. Any of these methods if isolated will produce nothing. Reality is like the ocean, infinite. It contains infinite information. And this is pretty much true, not an analogy or even a "perception". There's a correlation between the physical property called entropy and the concept called information entropy which measures the amount of information contained in a system. But as we are finite entities, we cannot process an infinite amount of data. If someone wants to drink ocean water, a cup is needed, whose volume is fixed (like a well defined concept). If not fixed (open hand, for example) water will leak. Ti is necessary, then.
As commented, using repeatedly Ti will imply the rules of logic. More than this: Maths work (I know about Gödel, but that's another issue) and they are, in certain way, the static essence of Nature. the always working generalization (Ti users love universals, right?). Let's think about numbers, we can have two pencils, two rubbers, two computers... two whatever, particular cases. If we supress particularization as much as we can, we would have simply "two". The idea of two is always there, always working, and always the same...
But we must never forget that logic rules work in this way: a logical deduction allows to us to deduce a true conclusion from a true premise. It says nothing about the validity of its premise, which has to be deduced form another premise and another and another... And if we don't cheat (circular reasonings) the first premise is just that, Reality or Nature (Spinoza's God). A glass of water is not the ocean, thinking they are equivalent is a reification fallacy, (overtrusting concepts) which is a fault Ti users have an higher chance of making. They tend (more than the rest) to use premises without questioning if they are valid, and if so, how much. And as Russell said: "with a false premise, you can prove anything". Please nobody get offended; I'm saying just Ti users have an higher chance for unconsciously making such mistakes. Once aware about this, this inclination could be corrected. The same applies to any type. Russell was likely LII, by the way.
Ti is still needed. In fact, pure dynamic conceptualization is pretty much an oxymoron. If some information-processing machine like human brain operates in such dynamic way that tracks external world at the same speed it changes, then there's no way to be aware about what's happening. Storing information (memory) requires a (at least) temporal disconnection from the source of information; if not, there's no storing, only "transparent" processing. For comparing new information with old one, it must be previously stored in certain way. Certain coherence ("definition") is required, which is not possible if all changes and nothing is preserved.
So Ti without Te is blind; Te without Ti is dumb. Both (or equivalent) are required, hence the title.