Information about how hedge funds work - like information about say how ad tech firms work - is incommensurable with HN's Reddit/Wiki-style way of disseminating knowledge. I hope that information without citations could be tolerated more, I have no dog in this race, I just like sharing knowledge.
Shaw and many hedge funds that have been around for decades acquire smaller arbitrage firms that can be thought of as having "licenses" to perform arbitrage. This portfolio of arb is how they scale to their AUM.
There is nothing actually secret or protectable about what an individual arb firm does to execute the trade. For example a particular HFT trading algorithm. Such firms that rely on actual secrets do not last long and so in a Darwinian way do not wind up in the portfolios of huge funds like DE Shaw.
The arb shops that last a while have permission, from a bank, that depends on a relationship with a real human being, that is exclusive to the firm granted, to launder a particular form of arb the bank would like to do but could not.
For example you're allowed to take a spread as a bond trading desk, but wouldn't you like to sell the bank's bonds instead of its competitors bonds? That's plainly against fiduciary so it "doesn't happen." Instead you go to an anonymous arb firm, for $1m in bonds, and you say, "okay we're going to sell you the bond that is actually ours, you will sell it back to us, here is your fee, now we can sell this bond back to the client and wash the fact that it is ours." Nobody says that! But that is the economics of the trade, why the arb firm can make money for so long, why banks work with them despite seeming to be "parasites," etc. The computers are just part of the hocus pocus of laundering that arb.
Shaw doesn't actually come up with this arb or necessarily source the relationship. Instead it finds the little anonymous arb firm and buys it, and carefully scales the relationship for $1m of bonds with 1 bank to $1b in bonds with 10 banks. The risks in Shaw's business are that you lose the license to arb. The upside is from investors persistently underpricing the ability for Shaw to scale this arb from $1m to $1b. No, they basically pull a rabbit out of a hat every year and continue to scale a particular arb further than people thought was possible.
Also, nobody is going to want to just give Shaw, a bunch of rich people, a bunch of free money. But you can't fight their ability to acquire little arb firms. This is the way.