In a quantum computer the hardware and software are intrinsically connected because the quantum Hamiltonian (or more precisely its time development) is the code that runs the computer. We demonstrate this subtle and crucial relationship by considering the example of electron-spin-based solid state quantum computer in semiconductor quantum dots. We show that multielectron quantum dots with one valence electron in the outermost shell do not behave simply as an effective single spin system unless special conditions are satisfied. Our work compellingly demonstrates that a delicate synergy between theory and experiment (between software and hardware) is essential for constructing a quantum computer.PACS numbers: 03.67. Lx, 85.30.Vw Ever since the pioneering work on quantum computation and quantum error correction [1][2][3][4] To help overcome these difficulties, more theoretical work is needed to explore the optimal operating regimes, figure out the operational constraints and tolerances, and discover potential sources of errors, just to name a few directions [15][16][17][18]. While the optical and atomic physics based architectures have been crucial in demonstrating the proof of principle for quantum computation, it is generally believed that solid state QC architectures, with their obvious advantage of controllable scale-up possibilities, offer the most promising potential for realistic large scale QC hardwares. The fundamental problem plaguing the solid state QC architectures has been the fact that the basic quantum bit (qubit), the QC building block, has not been compellingly demonstrated in any solid state QC architectures, although there is no reason to doubt that they exist in nature. Thus, the construction of successful QC hardwares has faced the somewhat embarrassing dichotomy: the architectures (ion traps, etc.) demonstrating existence of quantum bits cannot be easily scaled up, while the architectures (solid state QCs) which may be easily scaled up have not yet experimentally demonstrated quantum bits! Quantum computation with fermionic spins is considered to be a potentially promising prospect for solid state quantum computers [9][10][11][12]19]. Among the many proposed solid state QC architectures the spin quantum computer has several intrinsic advantages: (1) A fermionic spin, being a quantum two-level system, is a natural qubit with its spin up and down states; (2) it is fairly straightforward to carry out single-qubit operations on spin up and down levels by applying suitable magnetic fields (or through a purely exchange-based scheme [15]); (3) two-qubit operations can, in principle, be carried out rather easily (in theory, at least) by using the exchange interaction between two neighboring spins; (4) quantum spin is fairly robust and does not decohere easily (typical electron spin relaxation times in solids are many orders of magnitude longer [20] than the momentum relaxation time)-in particular, electron spin relaxation times could be microseconds in semiconductors [21].Our work presented in this...