Paradigm of the El Farol bar for modeling bounded rationality is undertaken. The memory horizon available to the agents and the selection criteria they utilize for the prediction algorithm are the two essential variables for agent strategies. The latter is enriched by including various rewarding schemes during decision making. Playing with the essential variables, one can maneuver the overall outcome between the comfort level and the endogenously identified limiting state. The distribution of algorithm clusters varies considerably for short memory. This affects the long-term aggregated dynamics of attendances. A transition occurs in the attendance distribution at the critical memory where the correlations of the attendance deviations take longer time to decay. A larger part of the crowd becomes more comfortable while the rest of the bar-goers still feel congestion for long memories. Introducing direct local interactions within the attendees by forming different types of networks, we create extremes in the attendance distributions. Delayed feeding of the data to the agents or their inclination of to absorb failure by insisting on unsuccessful algorithms introduces significant correlations, hence predictability of attendances. We additionally explore the extent of agentsí manipulation, achieved by modulating the threshold in accordance with the correlations in the data.