Our best guess for the structure of spacetime at the present time, is that is approximately of the form, X cross Y, where X is four dimensional Minkowski space, and Y  is a six or seven dimensional internal space. The geometry of the internal space, will determine the effective particle physics theory at low energy.tHe metric of Y will be Ricci flat at tree level, and will depend on a finite number of parameters, or moduli. However, one would expect quantum corrections and super symmetry breaking, to remove the degeneracy, and introduce an effective potential, V, which was a function on the moduli space of Y. The potential, V, could have a large number of local minima, corresponding to a landscape of possible vacuum states of M theory.sO what is it that determines that we are in the standard model state, and not one of the possible alternative vacuum states.  To answer this, one to turn to cosmology.oNe idea that has been advanced, is the eternal inflation scenario.iN this, quantum fluctuations in a scalar field, phi, are supposed to drive the potential, V, up to the Planck value. The universe as a whole would expand in an inflationary manner.hOwever, in some regions, quantum fluctuations would reduce the potential below the Planck value, and cause phi to start rolling slowly down the potential hill. Widely separated regions, were supposed to fall into different local minima of the potential, which would give the universe a mosaic structure, with different parts of the universe, in different vacua.Eternal inflation is an essentially classical picture, which assumes there is a single metric for the universe. tHat is why its advocates feel it is necessary to suppose the universe has a mosaic structure, to accomodate the possibility that the universe could be in any of the vacuua. However in a fully quantum theory, the universe can have any metric with suitable boundary conditions, which I shall take to to be the no boundary condition. The amplitude for a particular vacuum, is then given by the path integral over all no boundary metrics with that final state. One can also calculate the amplitude for inhomogeneous final states which are mosaic of different vacuua.tHey will in general be lower than the amplitudes for for homogeneous final states.One can estimate the amplitude for a particular vacuum,  by finding a no boundary solution which ends in that vacuum.tHe amplitude is then  the exponential of minus the action of a classical solution, times a prefactor, which I shall ignore. One no boundary solution would be de Sitter space, with the value of the vacuum energy corresponding to the local minimum of the potential. The action will be large and negative, so the amplitude will be high, particularly if the vacuum energy is very low. However, this solution is completely empty, and therefore not of much interest.tO obtain a universe like the one we live in, with matter that is clustered into galaxies and stars, it seems necessary to have a period of inflation. Inflation will occur in solutions in which the potential is initially  high, and rolls down the hill to one of the local minima.bUt why should the potential be high initially. We have already seen that solutions which start at or near a minimum, have a high amplitude, but no matter content. On the other hand, a solution that starts at a maximum or a saddle point, will have a much lower amplitude, but will have a long period of inflation, and can produce a matter filled universe, like the one we live in. One might think that solutions that started lower on the potential hill, would have a much larger amplitude.tHis would result in the most likely inflationary universe, having a very short period of inflation, and a large spatial curvature today. However, Thomas Hertog and I showed that if the maximum or saddle point was broad, the solution which started at the top would have a higher amplitude than solutions starting nearbye. There are several important consequences of this result.First, if there is a broad saddle point below the Planck density, the dominant contribution to the path integral will be from metrics in the semi classical regime.tHe universe will undergo a bounce at finite density, and there will be no singularity. The success of the singularity theorems of Penrose and myself, led everyone to assume that the universe began with a singularity. However, the singularity theorem applicable to cosmology, though not that for black holes, depends on the strong energy condition. This is satisfied by Maxwell and Yang Mills fields, but not by scalar fields.It is therefore possible to have a completely non singular solution, that collapses from large size, bounces at finite density, and expands again. Such non singular bouncing solutions, form only a small sub set of the space of all Einstein scalar field solutions, but it is they that are selected by the no boundary condition. They dominate the amplitude for the present state, because of their large negative action. Solutions in which the potential reaches the Planck value, will have only a small action, and can be neglected. The origin of the universe, is in the low energy regime of M theory, in which four dimensional general relativity, is a good approximation. This is supported by the fact that calculations based on four dimensional general relativity, are in excellent agreement with observations of the microwave background. One would not expect this, if the four dimensional approximation, X cross Y, broke down before one gets back to the time of inflation. This would indicate that the internal space, Y, is smaller than 10 to the 6, times the Planck length. The only situation in which 4d general relativity would break down, and in which one would need string theory, or some other approach, would be the final stages of evaporation of a black hole.Coming back to the landscape, the only vacuum states that will have significant amplitudes, will be those where the minimum of the potential, lies on the line of descent from a broad saddle point of index one. By this I mean that there is only one direction in which the action decreases. This is in accord with the general principle, that the instanton that describes tunneling from one vacuum to another, should have one, and only one, negative mode. In this case, the instanton would be the Hawking Moss or de Sitter instanton, with the value of the vacuum energy at the saddle point. The negative mode, would be the homogeneous mode in which the scalar fields everywhere move along the line of steepest descent from the saddle point. If the saddle point is broad, that is, if V double prime over V, is small and negative, the lowest inhomogeneous will be positive, giving the Hawking Moss instanton one, and only one, negative mode, as advertised. One might interpret the instanton, as describing the tunneling from the vacuum on one side of the saddle point, to the vacuum on the other side. But it is not tunneling between empty universes.The slow roll down from the saddle point, means that the universe is filled with matter, and therefore a possible candidate for the universe we live in. Many of the saddle points will not be broad enough, that only the homogeneous mode, is negative.Such saddle points will not give an amplitude for inflation. Thus only some of the states in the landscape, will have a significant probability. This would go some way to satisfy those physicists who had hoped to predict the low energy effective particle physics theory. But it couldn't restrict it to just a single state, because each broad saddle point, would give amplitudes for the minima on either side of it, and there are likely to be a number of broad saddle points. So one will still need to appeal to the anthropic principle. Indeed, the principle has already been used in restricting attention only only matter filled universes, and ignoring the empty de Sitter universes at the minima of the potential, which will have much higher amplitudes. The anthropic principle, in some form, is unavoidable.In the usual, bottom up, approach, one assumes that the universe started in a state of high symmetry, which then this evolved the present broken symmetry state. The symmetry breaking would happen in different directions in different places, leading to topological defects, such as domain walls, cosmic strings, and monopoles. On the other hand, according the top down approach I have described, the solution that gives the dominant contribution to the amplitude, could have the same broken symmetry all the way back. In this case, there would no production of topological defects.The no boundary condition, enables us in principle to calculate quantum amplitudes, for the whole universe to be in different states in the landscape. It would be a mistake to assume that we should be in the state with the highest amplitude. That would be like saying I should be Chinese, because there are many more Chinese than Brits. If the amplitude for the standard model vacuum is non zero, it is is irrelevant what the amplitudes for other vacuum states are. The theory can not predict a unique vacuum state.iNstead, we have to input that we live in the standard model vacuum.To sum up.tHe no boundary proposal, allow us to calculate amplitudes for matter filled states in the landscape. The dominant histories in the path integral for these amplitudes, lie entirely in the semi classical regime of four dimensional general relativity. The amplitudes will be highest for states in which the whole universe is in a single state, rather than a mosaic of different states, as predicted by eternal inflation. There will be no primordial production of topological defects, such as monopoles, and cosmic strings. Not all states in the landscape will have significant amplitudes, but there will be more than one that do, so M theory does not predict a unique low energy particle physics theory. But we have already used anthropic arguments in restricting attention to matter filled universes.tHe amplitude for empty de Sitter space, is much higher. The theory predicts a set of amplitudes for matter filled landscape states.iT is implausible that life is possible only in one of them, so we might have chosen a better location.