In a new kind of prediction network, self-navigating Python, R and Julia algorithms conspire to produce superior electicity predictions than the official forecasts - then automatically review model residuals. They also find their way to any published time series, thereby providing essentially free prediction to anyone who needs it. I will discuss the potential for collective real-time prediction, and demonstrate a prototypical host at Microprediction.Org. Parts of contest theory and a lottery paradox are highly relevant to algorithms submitting distributional predictions.