I saw it here: www.aps.org/funding-reco...

The APS did post that the 2026 winners were updated (on twitter) along with a couple press-releases.

This issue has been known for a long time:
arxiv.org/abs/1701.08750
arxiv.org/abs/1709.07882
(figures from these papers)

If the goal is to make an actual discovery (not just get new bounds into PRL), one needs to be honest about what we know.

That said, we have learned a lot in the past year and we should have a paper out soon that will better explain the situation.

Forecasts tend to saturate at the optical depth limit post DESI. So there isn’t huge statistics gain coming. If it is just a large statistical fluctuation, there isn’t a guaranteed way to figure that out.

The answer is a bit complicated. The short answer is that you will gain a bit more going from current to full DESI but after that there really isn’t anything planned that will help. Much more likely to find systematics to explain such a change.

Sorry - if you look at the paper I referenced, they let the optical depth be a free parameter without including the low ell data that usually fixes it. They find both the expansion history and neutrino mass look fine, but the optical depth is roughly what we estimated in our paper (6 sigma high)

(3) The optical depth can improve the whole situation but indeed as high as suggested (arxiv.org/abs/2504.16932) (4) this isn’t inconsistent because the current error is not completely determined by tau (yet)

There are a few things going on: (1) yes, when you allow neutrino mass to be negative, the exclusion of the the physical region consistent with oscillations get stronger (99% in our paper). (2) this does depend on how you define (negative mass), so not all the papers agree on the exact value.

In fact, the reason the SPT result is lower than previous ones is that they use a different value for the optical depth (there are multiple values that come from Planck that are just slightly different ways of analyzing the low-ell data). The SPT data isn’t actually driving this.

What you would need is that there is a systematic error, most likely in the optical depth that is 6-7 sigma. You could tell by repeating this measurement. Unfortunately you will need another CMB satellite to check, which is 10-15 years away (if it happens at all)

The logic of “more data” is the it is a statistical fluctuation that will be revealed my just adding more data. That is literally not possible. You can add ever increasing quality CMB and BAO data and it will do nothing to the neutrino mass measurement.

The really critical point is that “more data” is not a solution. Because of the peculiar nature of the neutrino mass measurement, it is limited by measurements that will be very difficult to repeat and may not happen again for 15 years

There is very little about the SPT result that is new - this issue with neutrino mass has been there since DESI Y1 BAO. An initial run-down of the possible explanations is here

arxiv.org/abs/2405.00836

I think it is also the problem that deep insights into nature happen slowly. Often there are pieces that fall into place over years so by the time we agree something is true it also feels like old news. Unfortunately, Nobel prize announcements may be the only time we get to really celebrate them.

There are shifts in the value from reanalysis, but nothing big enough to matter. I also haven’t seen a believable claim that a 6-7 sigma error in tau (needed for neutrinos) would definitely have no impact on the DE interpretation of DESI+CMB.

SPT crucially uses polarization only so it has completely different systematics (and data) from the other two

That is not true. The issue with the mass is consistent with Planck, ACT and SPT.

If that is how we should think about your data, then your are guaranteed only to find DE. In DR1, rather than worry about the negative neutrino mass, the collaboration added a prior that it is >60 meV to make it look like nothing.

The main point I am making is that DESI does not explore many alternatives to the DE interpretation so we shouldn’t just assume that is even close to the best explanation. It is convenient to say we shouldn’t trust the other analyses because they collaboration didn’t do it.