In his book A New Kind of Science, Stephen Wolfram described a universal 2-state 5-symbol Turing machine, and conjectured that a particular 2-state 3-symbol Turing machine (hereinafter (2,3) Turing machine) might be universal as well.
On May 14, 2007, Wolfram announced a $25,000 prize to be won by the first person to prove or disprove the universality of the (2,3) Turing machine. On 24 October 2007, it was announced that the prize had been won by Alex Smith, a student in electronics and computing at the University of Birmingham, for his proof that it was universal. Since the proof applies to a non-standard Turing machine model which allows infinite, non-periodic initial configurations, it is categorized by some as "weak-universal".
Background
Claude Shannon first explicitly posed the question of finding the smallest possible universal Turing machine in 1956. He showed that two symbols were sufficient so long as enough states were used (or vice versa), and that it was always possible to exchange states for symbols.
The following table indicates the actions to be performed by the Turing machine depending on whether its current state is A or B, and the symbol currently being read is 0, 1 or 2. The table entries indicate the symbol to be printed, the direction in which the tape head is to move, and the subsequent state of the machine.
{ | class="wikitable" ! style="width:20px;"| !! A !! B |- ! 0 | P1,R,B || P2,L,A |- ! 1 | P2,L,A || P2,R,B |- ! 2 | P1,L,A || P0,R,A |}
The (2,3) Turing machine: Has no halt state; Is trivially related to 23 other machines by interchange of states, symbols and directions.
Minsky (1967) briefly argued that standard (2,2) machines cannot be universal and M. Margenstern (2010) provided a mathematical proof based on a result by L. Pavlotskaya in 1973 (not published but mentioned in Margenstern article); thus, it might seem that the (2,3) Turing machine would be the smallest possible universal Turing machine (in terms of number of states times number of symbols). However, the results are not directly comparable, because Minsky only considers machines that explicitly halt, which the (2,3) machine does not. More generally, almost all formal definitions of Turing machines differ in details irrelevant to their power, but perhaps relevant to what can be expressed using a given number of states and symbols; there is no single standard formal definition. The (2,3) Turing machine also requires an infinite non-repeating input, again making a direct comparison to earlier small universal Turing machines problematic.
Therefore, though it may be true that the (2,3) machine is in some sense the "smallest possible universal Turing machine", this has not been strictly proven in the classical sense, and the claim is open to debate when taking into consideration traditional definitions of universality and whether the relaxing of the Turing machine properties used for the proof can be allowed in general and may even suggest novel ways to define computational universality more independent of arbitrary choices (such as having a halting configuration or requiring a blank tape).
The state of the head (up or down droplet (A and B respectively)) and the pattern of color (white, yellow and orange (0,1, and 2 respectively)) in a given row depends solely on the content of the row immediately above it.
Even though the machine has a head with only two states, and a tape that can hold only three colors (depending on the initial content of the tape), the machine's output can still be arbitrarily complex.
Proof of universality
On 24 October 2007, it was announced by Wolfram Research that Alex Smith, a student in electronics and computing at the University of Birmingham (UK), proved that the (2,3) Turing machine is universal and thus won Wolfram's prize described above.
The proof showed that the machine is equivalent to a variant of a tag system already known to be universal. Smith first constructed a sequence of rule systems showing that the (2,3) Turing machine is capable of arbitrary finite computations. He then employed a novel approach to extend that construction to unbounded computations. The proof proceeds in two stages. The first part emulates the finite evolution of any two-color cyclic tag system. The emulation is a composite of a series of emulations involving the indexed rule systems 'system 0' through 'system 5'. Each rule system emulates the next one in the sequence. Smith then showed that even though the initial condition of the (2,3) Turing machine is not repetitive, the construction of that initial condition is not universal. Hence the (2,3) Turing machine is universal.