To support the immense traffic growth and enable scalable on-demand provisioning of service requests, optical networks must deliver great agility and reconfigurability in a cost-and resource-efficient manner. The progress in elastic coherent transmission [1] has enabled simplifications in the optical network architecture by, for example, replacing the costly wavelength switches in the add/drop part of colorless reconfigurable add/drop multiplexers (ROADMs) with passive couplers. The concept of passive filterless networking further proposes the elimination of wavelength switching in the transport function as well. It is based on completely passive interconnections realized by passive splitters and combiners, essentially forming fiber trees spanning the network nodes [2]. In such networks, transmission follows the Drop&Waste ( or Drop&Continue) scheme where each signal is broadcasted to all branches downstream of the source node in a tree, while a copy of signal continues to propagate downstream of the destination node [3]. Thus, the simplification of the nodal architecture comes at the expense of increased spectrum usage due to the presence of unfiltered signals. Moreover, the broadcasting of signals to inadvertent nodes raises confidentiality issues. An additional drawback stems from the inability of the passive, fixed interconnection of nodes to allow for topology reconfiguration. The optical white boxes, or programmable optical switches (also referred to as architecture-on-demand) can provide unparalleled switching and architectural flexibility [4]. Unlike in conventional ROADMs, optical modules (e.g., passive couplers, amplifiers or WSSs) inside an optical white box are not interconnected in a hard-wired manner, but are selected on demand by a reconfigurable optical backplane (OB), implemented by, e.g., a piezoelectric switch or 3D MEMs. An arbitrary nodal architecture can be configured by setting up the interconnections of the OB as per traffic requirements, and swiftly reconfigured to accommodate changing traffic demands, scale capacity, and ease migration and upgrade. Combining the agile filterless transmission with flexible optical white boxes into a programmable filterless network architecture, recently proposed in [5], integrates the agility of filterless operation with the high flexibility enabled by white boxes. The nodes in a programmable filterless network are equipped with a programmable OB hosting only passive couplers to route the connections. The preliminary study of routing, modulation format and spectrum assignment (RMSA) in programmable filterless networks aimed at spectrum usage minimization [5] shows that they are capable of reducing the amount of unfiltered signals compared to passive filterless networks and can significantly decrease spectrum usage. Due to the moderately-sized switches deployed for the OB, this architecture also offers the potential of diminishing the nodal costs compared to the conventional ROADM-based networks. In addition, when striving to minimize the spectrum usage, connections tend to get routed over fewer splitters as they are the cause of broadcasted, unfiltered signals. This may lead to a decrease in the total number and degree of passive couplers traversed by connections, which can reduce the signal losses and lower the amplification requirements. This talk will outline the operational principles of programmable filterless network architecture and their benefits in terms of spectrum and component usage. We will also present possible technological and optimization approaches to further enhance the performance of white box based filterless networks and the related preliminary results.