I'm wondering, what incentive do researchers have to share their new compression algorithms with JPEG? Especially if JPEG patents everything as you suggest.
If it works similarly to other standards groups, they don't take the patents from inventors. Instead, they manage the patents and collect fees on behalf of the (many) holders, allowing anyone to buy the rights through a single entity. The various MPEGs and other formats like DIVX work in this way.
No that isn't true. ISO (MPEG/JPEG) does not play a part in licensing patents or forming patent pools, which is only natural when you consider the fact that ISO is an international organization headquartered in Geneva and every country has its own patent ecosystem.
Individual companies within a certain jurisdiction can certainly set up pools and court the various rightsholders who participate in the standards process, and that is what you see with e.g. MPEG-LA and HEVC Advance in the US. But these have no direct connection to ISO beyond using the name and serving the contributing organizations.
Also, JPEG and MPEG are both part of the same standards group. And Divx is just a brand of MPEG implementations.
MPEG-LA (the patent pool organizer) isn't MPEG (the standard body), even if the former adopted the latter's name.
The only thing to blame standard bodies for is their dysfunctional approach at patents that incentivizes useless feature stuffing so that all parties have some patents in the mandatory-to-implement part of the spec.
In the U.S., you have one year from the initial date of public disclosure to file a patent application, so you can still patent it after publishing a paper, if you don't wait too long. See 35 U.S. § 102(b)(1), https://www.law.cornell.edu/uscode/text/35/102
