Lottery payouts are structured to ensure the state always comes out ahead. The state takes its share off the top, presumably along with the vendor fees; the jackpot is what's left over. It's not stated plainly, but the article does mention it:

> The Texas Lottery Commission heralded the win, the third largest in state history, which helped raise around $50 million for the state’s public schools out of $138 million in sales over the life of the jackpot run.

That doesn't quite add up as the the jackpot was $95 million ($138 - $50 == $88), so perhaps the article overstates the net revenue or got some numbers wrong?

To the extent anyone lost something concrete, it was those Texans who play the lottery; their expected payout was smaller than it might otherwise have been as they were guaranteed to have to split the pot with the syndicate. OTOH, strictly speaking the syndicate also took a commensurate risk. If 3 or possibly even 2 other players had also won, they would have lost money.

But there's also the loss of confidence and injury to people's sense of fairness, where mathematical odds are only one part of the fuzzy equation. The Texas Lottery organizers thought they were doing right by the state. The more tickets sold, the more money in net revenue. But maybe they should have considered more the long-term implications to the lottery's image and stable revenue streams. Though, at the end of the day perhaps they still made the right decision as fiduciaries, notwithstanding some of them seem to have lost their jobs in the process. In similar cases in other states (usually involving scratch offs?), lotteries knew for years about net-positive schemes, but kept the odds structures as they brought in greater revenue (e.g. attracting interest from savvy players outside the state) regardless. Because the vast majority of players don't respond perfectly elastically to expected odds, especially when they're not a fixed function (e.g. they don't consider things like jackpot sharing), it's arguably a legitimate approach to increase revenue, at least until the schemes become well known.

If each ticket was $1 and then there are 21 million unique numbers, then on average there are 6.5 people per draw.

Also in any case someone buying one of everything vs buying randomly the same amount wouldn't affect the expected return of anyone.