I'm moving this discussion here, per Neven's observation that it was off-topic in the "2018 melt season" thread:
I think a large part of the issue here is that most scientists acknowledge that they cannot be sure how complex systems will respond to stresses in the future, and therefore they speak and write in terms of probability. Thus Archimid states there is a chance, but only a very low chance that the Arctic would melt out this year, and that the probability increases in future years as the Earth (probably) warms up. It really is the only sensible way to speak of the future. If we toss ten pennies, it is unlikely that all ten will come up heads, but it is also possible, and we can assign a (low) numerical probability to that.
It is difficult to predict possible outcomes in the case of huge natural systems like the Arctic, but we can ascribe probabilities based upon trends, algorithms, computer models etc. Then the experts argue about it all, gather more evidence and recalibrate their predictions, which is how it should be in science. And we continue to be surprised, and often humbled, by how the planet and the Arctic responds.
No chance of an ice-free Arctic for at least 20 years.
That is false. There is a very low chance this year, increasing every year that gets warmer and the weather more extreme. Saying there is no chance it will happen before 2037 is simply not true.
This year the uncertainties are the early opening of the Bering, a thin CAB, no ice North of Svalbard and the ever present warmer planet.
What if I said 15 years instead? Would that satisfy you?
There really is no chance of an ice-free Arctic for at least 15 years, but probably more like 20 years. The ice extent will continue to extend in winter, and will continue to bulge up and thicken against the CAA and Greenland. That the whole thing could be less than 15% in a few years is unlikely.
Much smaller ice-pack bunching up? Yes.
Zero ice-pack anywhere, and just some icebergs distantly floating around? No chance.
Here's one way to think about this. First, here's the historical record of JAXA annual (daily) minima:
The blue zigzag line are the actual observations, the green smooth line is a LOESS smoothing function with a 10-year timescale, the red smooth line is a 30-year LOESS function, and the blue dot with error bars is the current forecast for 2018.
That graph suggests a variety of possible future scenarios. Here are four examples or "scenarios". The first scenario is based on the idea that the 10-year LOESS model (green line above) shows a "stepwise" decrease, where extent drops sharply for a few years, then levels off for a few years, then drops again:
In that scenario, the ice could be around a lot longer than most of us expect -- potentially, there might not be an "ice-free" day (let alone month or summer) until 2080 or later.
If we focus instead on the red line (30-year LOESS) from the first graph, it tells a different story -- a slow decrease in the 1980s and 1990s, followed by a slightly faster decrease after 2000, with the apparent "flattening" over the past decade being just noise. In my scenario 2 below, the post-2000 decrease continues unabated:
This scenario has an "ice-free" day sometime around 2035-2055.
A third scenario assumes that in 2020 we have another downward "bend" like in 2000, followed by a steeper loss of ice:
That suggests an "ice-free" day somewhat earlier, between 2030-2040.
Finally, a fourth scenario, that basically ignores the historical record and assumes that the ice suddenly collapses after 2020:
Personally, I think scenario 3 is most likely. I would expect to see "ice-free" days in September routinely by the mid-2030s, with the first one perhaps occurring only 10-15 years from now given the apparent increase in interannual variability.
After that, I would rate scenario 2 as next-most likely, with scenarios 1 and 4 as potentially plausible but less likely.
FWIW ... I put "ice-free" in quotes throughout this post, because I personally dislike the standard definition of "ice-free" as under 1 million km2 of ice. See this comment:
https://forum.arctic-sea-ice.net/index.php/topic,143.msg122338.html#msg122338and this map (another simulation):