This is not about exploiting a privileged access to market conditions (as in the sub millisecond access HFT firms have) in order to gain an advantage.
This is about using machine learning to predict future (as in hours/days) moves on the market and take advantage of them.
Speed is nice, but smarts are better. "Machine Learning" is the new buzzword for a way to mathematically deducing future outcomes based on data science and quantitative analysis. That is a grotesque simplifications and there are specific ways of approaching this (Tensorflow/Keras from Google, Torch from Facebook, etc). This could be nanoseconds in the futures, it could be hours, or even weeks / months. And this is the part you're failing to understand by trying to assume (incorrectly) that all HFT is made up of is bid sniping and latency arbitrage (often referred to as front running). It isn't really that privileged access if you yourself can pay any exchange, such as NYSE, for a feed to your house.
TL;DNR: Your understanding of electronic / HFT trading is wrong. If you try to understand it first, you'll realize your statements bear absolutely no basis whatsoever in facts. Facts are that Electronic Trading firms have been using machine learning literally since the invention of machine learning and quantitative analysis to make money. A new firm trying to do this can simply join the club.
Electronic trading as opposed to what?
And even if I call my broker, and he clicks somewhere to enter the order, it's still executed by some HFT agent somewhere. And the same applies to low frequency strategies, I guess (why bother with the execution details, just use a broker).
TL;DNR: You keep ignoring both the point of the comments and the point of the TFA. This is about market prediction, not technical exploitation of a privileged access to market.
anyone can buy co-located rack space.
If you buy a co-located rack space you are using a PRIVILEGED CONNECTION to the market that and taking advantage of that using some algorithm: That's High Frequency Trading.
The all point of TFA and the comments was the OPPOSITE: To be in a NON PRIVILEGED position (as in, the same access time to market transactions as the median of the access time to market of EVERYONE ELSE) and to use machine learning to predict market moves in a long enough term so that the market access time was irrelevant.
Do you understand the complete difference between the two scenarios now?