michal_dubrawski

Michał Dubrawski
About:
Show more

No Scores Yet

Relative Brier Score

14

Forecasts

62

Upvotes
Forecasting Activity
Forecasting Calendar
 

Past Week Past Month Past Year This Season All Time
Forecasts 1 7 77 14 146
Comments 2 12 180 30 369
Questions Forecasted 1 4 8 4 15
Upvotes on Comments By This User 4 27 411 62 670
 Definitions
New Prediction

I am updating slightly based on the reply which @ctsats got from the head of the HWC, Prof. Abel Méndez (https://twitter.com/ProfAbelMendez) (Impresive work Christos! Thank you for sharing!) Especially professor's guess of the number of planets added this year (as 5 to 10). We must however note that if my previous analysis is right (see my previous forecasts here) 7 exoplanets were added to the catalog this year (6 from 2023 and one from 2015), as I explained here, professor likely counted some of them as if they were added the previous year - otherwise he would say 7 to 10, right? So it is hard to interpret how many (if any) he already counted for this year, and our period of analysis is different, it doesn't include January this year, but in fact includes January 2025 (despite 31st of December 2024 being mentioned in the question title). Still, we have a full-year period that way and unless this high assessment from the professor if heavily based on the planets already added to the catalog this year before the question was published and not counting towards the resolution, this assessment is still useful information.   

Files
New Prediction

Reducing slightly. Before I dig deeper into the WayBack Machine it is hard to evaluate how likely are new additions from the previous year, and what is the real cause behind these additions (I wrote about some possibilities I came up with, below).

I still wonder why they added the one from 2015 this year - they already have potentially habitable exoplanets  from all years since 2013 in the catalog, and we have 8 from 2015 (5.1% of all 875 exoplanets discovered that year), while we also have 8 for 2014 - only 0.9% from all 875 exoplanets discovered that year. 2016 also has a low share of habitable worlds in comparison to 2015 - only 11 from 1517 (0.7%). I wonder what is the cause behind these differences. I still need to understand the process of evaluating habitability of the exoplanet better - maybe for some methods we don't have enough data to judge if the planet is potentially habitable if other observations are not made? Maybe some new information about K2-3 d changed the previous assessment of its habitability? Last update in P_UPDATE column for this exoplanet is from 10.07.2023, so that would make sense. I must read more about this exoplanet.
Also do they have three values: 1) potentially habitable, 2) not habitable, and 3) not known/not enough data? Or maybe the research team behind the catalog is still evaluating some of the exoplanets found in the past? My intuition tells me that this is more likely because of new data, like about new information about the water world ("LHS 1140 b is a potentially habitable water world").

Also, I wonder about the candidates awaiting confirmation - if confirmed, would they be assigned as a year of being discovered to the year of first spotting (when they become candidates) or the year of confirmation? There is so much I do still not know about all this.

Tagging @Tolga @martinsluis @belikewater @404_NOT_FOUND @ctsats @cmeinel @probahilliby @PeterStamp @ScottEastman @NoUsernameSelected @DimaKlenchin @WeirdAwkward @guyrecord @Perspectus @JJMLP  

Files
michal_dubrawski
made a comment:
Great point @ctsats! Thank you. I will write a clarification request soon, and can also write about this on our Discord. 
Files
New Badge
michal_dubrawski
earned a new badge:

Active Forecaster

New Badge
michal_dubrawski
earned a new badge:

Star Commenter - Mar 2024

Earned for making 5+ comments in a month (rationales not included).
New Prediction
Why do you think you're right?

By finding the url address of the previous version of the catalog website and by analyzing archived versions of both websites (for details, please see my a bit chaotic rationale here and comments to it, especially this and this) , I was able to on the previous  was able to establish that this year there were 7 exoplanets added to the catalog as potentially habitable, including one added on March 21 after the question has started (it is TOI-904 c - publication about it is from December 2023 - do not to confuse it with LHS 1140 which was already in the catalog - see this comment for details), so this one counts towards 5 needed for the resolution as "yes".

So the planets added this year are all 6 discovered in 2023:
TOI-700 e 

TOI-715 b

HN Lib b

Wolf 1069 b

GJ 367 d

TOI-904 c (added to the catalog on the 21st of March, so it counts toward the resolution, but it shows that some potentially habitable exoplanets discovered in the late 2024 might not be added to the catalog in time to be counted for the resolution, so this kind of nullifies advantage we got from TOI-904 c being counted late)

The last exoplanet added to the list this year is K2-3 d and what is critical information, it was discovered in 2015 (see this comment for how I established it). So now we know that the researchers behind the catalog are not only adding the most recent discoveries but are also analyzing exoplanets discovered in the past (see this chart) - without further analysis of archived data it is hard to say how often they find something they have not yet found in the previous years, but 2014 and 2016 gave them a lot of material, and they might be still analyzing some of it. So there is a chance that this will be a source of some additions to the catalog this year (or in January 2024) as well.

Tagging @404_NOT_FOUND @ctsats @cmeinel @probahilliby @PeterStamp @ScottEastman @NoUsernameSelected @DimaKlenchin @WeirdAwkward @guyrecord @Perspectus @JJMLP 

Files
Why might you be wrong?

I need to understand their update pattern and research vs publication (update on website) timeline. Will they soon go dark until January 2024? 


Files
New Prediction
Why do you think you're right?

To paraphrase John F. Kennedy: The greater my knowledge increases, the more my previous ignorance unfolds.  It is great to be with a team with awesome thinkers like @ctsats and others here, who never stop digging deeper and challenging their own understanding of the matters we forecast about. Highly recommended comments by Christos are here and here.


*** UPDATE: 2024-03-30 - I was wrong about the P_UPDATE column in my rationale I presented below - see my comment below or follow the link. ***

So, the insight about Kepler no longer being active for years cannot be read strongly since as @ctsats explained, there are still 3353 candidates exoplanets found by Kepler which still wait for confirmation or disconfirmation, and about 5000 candidates in total.

Now, things get interesting because the discovery rate is not the same as the rate of updating the catalog on resolution website and as @ctsats pointed out the year of discovery is not equal to the year of exoplanet being added to the catalog.
The most recent update (March 21st) added TOI-904 c to the catalog. This exoplanet was discovered in 2023 (This publication listed here was published on December 7 2023), so we may think that there is a time lag of 3,5 months between publication of the article announcing the discovery and the planet being added to the catalog. But the problem is that as Christos mentioned, this is the third planet added to the catalog this year (in fact seventh if my analysis below are correct). We can confirm this with WayBack Archive for the copy of the website from the 9th of January this year: https://web.archive.org/web/20240109164551/https://phl.upr.edu/hwc


  So before the question was published they added two more planets - I initially assumed that the most recent ones were added to the top of the list that would give us: 
TOI-700 e

TOI-700 d

So the problem is that TOI-700 e was announced in this publication from February 16, 2023  https://iopscience.iop.org/article/10.3847/2041-8213/acb599 and reported in the news in March https://theconversation.com/distant-star-toi-700-has-two-potentially-habitable-planets-orbiting-it-making-it-an-excellent-candidate-in-the-search-for-life-198274 this suggest about a year delay, much longer than with the latest publication.

 And TOI-700 d was discovered on January 3, 2020 (Wikipedia specified the daily date, often I only see there a year: https://en.wikipedia.org/wiki/TOI-700_d) and the publication is from August 14, 2020: https://iopscience.iop.org/article/10.3847/1538-3881/aba4b2

I read this shortening intervals between the publication and addition to the catalog that the team behind the Habitable Worlds Catalog is themselves catching up to with the publications. So their three recent additions covers between 2020 and December 2023 but we now have 6 habitable or potentially habitable exoplanets discovered in 2023 in the catalog. This looks like they were adding both recent ones and some from the past. There is potential that they will add more exoplanets, which were discovered in the previous years but are not yet in the catalog. 
@ctsats I think we can learn more by the WayBack Machine on Archive.org, the trick is to know that the catalog has chaned its name and because of that also the website. I found the previous url by looking at Wikipedia editions history: https://en.wikipedia.org/w/index.php?title=List_of_potentially_habitable_exoplanets&oldid=1209717803 the previous url is https://phl.upr.edu/projects/habitable-exoplanets-catalog - if we work on that archived links we can learn how fast they were adding the exoplanets to the list and how often the list was updated (unfortunately interactive list of planets doesn't work in the recent versions of the website and many of the archived versions doesn't even have the number of planets visible - the total number is visible from December 2022). A year before the question opened there were 63 exoplanets on the list so they added 7 till now  https://web.archive.org/web/20230228020942/https://phl.upr.edu/projects/habitable-exoplanets-catalog but what is interesting the archived version from 31 December 2023 still shows 63!  https://web.archive.org/web/20231231171144/https://phl.upr.edu/projects/habitable-exoplanets-catalog - That would mean that they added 7 exoplanets to the catalog this year around the time of reorganization related to creating the new website in January this year.

What is interesting, between December 2021 and December 2022 the catalog was updated only once: https://web.archive.org/web/20221204232019/https://phl.upr.edu/projects/habitable-exoplanets-catalog - we can see 61 exoplanets here.
before that, we can see on version from December 2nd 2022  https://web.archive.org/web/20221202224039/https://phl.upr.edu/projects/habitable-exoplanets-catalog "Last Update: December 6, 2021" - but in the csv we can see that they have been doing more updates to the catalog that year (see my table below - not all updates mean that the exoplanet was added, at some cases information about previous exoplanet was updated).

We previously established that P_Year is a year of discovery, P_Update must be the last update date of any column. If based on the WayBack Machine archived version from 31 December 2023, we now know that 7 planets were added to the catalog this year, than we know that this must be some of the 8 with most recent update dates from this year, in the eight case I guess some other column was updated rather than it being added to the catalog. See my table below:

(sorry, one exoplanet from 2023 is not colored above - my LibreOffice Calc keep crushing and I lost the table)

This means that from the 7 added this year at most 2 were from 2023, and at least one was from 2013 (because there are two exoplanets from 2013 and at leat one must have been added this year).

If they keep up with this rate of public updates and if they will still be adding exoplanets discovered in previous years I think it makes it very probable, but I am not sure how they are organize -  maybe the current updates are based on the past months of work, and new updates may require similar months of work. At the same time, if the additions are mostly from the previous years that unfortunately might mean (depending on the Question Team intentions), that resolution criteria may not fit well to the question of new exoplanets being found to be habitable since these exoplanets might have been found to be habitable many years ago. I guess the scientists behind the project put their own work into the habitability evaluation process and not only copy the results from others work, so that still makes sense because it becomes less about discoveries of new exoplanets and more about discovering exoplanets habitability.      

Files
Why might you be wrong?
If this does not happen, I guess that should be the result of the team behind Habitable Worlds Catalog research (maybe they are all now publishing results from their past few months) and updating schedule (something like going back to work and not updating anymore like their one update in 2021, but I don't think it is probable, at the same time last year they were not publicly updating the catalog between February and December). It might also be the result of team catching up with the scientific publications from the previous years, but that seem less likely. 
Files
michal_dubrawski
made a comment:

@404_NOT_FOUND  @ctsats @cmeinel  We can get the estimate about when exoplanet was added to the conservative sample of potentially habitable worlds by analyzing the images on the archived version of previous website: 

For example:

the image from the January 2024:

From the 5th January 2023:

 

So these are the new ones:


TOI-700 e  was discovered in 2023

TOI-715 b was discovered in 2023

Wolf 1069 b was discovered in 2023

K2-3 d was discovered in 2015

So we were able to find 7th exoplanet added to the list in 2024 (the only one that is not from 2023) and it is K2-3 d from 2015. It is important information because the fact that it is not some exoplanet found in the late 2022 confirms that they can add exoplanets which were already found many years in the past but are now found by the researchers to be fitting to the potentially habitable world criteria. 

Files
New Prediction

I am going slightly down since I got this answer from the INFER team as a reply to my request for clarification about the cyberattack: "This question is concerned with a cyberattack that damages or interrupts the normal functioning of critical infrastructure as mentioned in the clarification, rather than theft of data/intelligence gathering that could lead to such harm in the future." So not all successful cyberattacks count, even if they were directed at the critical infrastructure (like data theft and probably also disclosing stolen information or publishing disinformation on the Iranian government websites which would create any public reactions affecting the critical infrastructure).

Gonjeshke Darande, the hackers group behind the previous attacks on Iran (including the one on gas stations from December last year) is worth watching. They proved to have the capabilities and motivation to do something like that, however they might choose some other kind of cyberattack - not necessary affecting critical infrastructure again. It looks like they have their profile on X: https://twitter.com/gonjeshkedarand?lang=en and a channel on the Telegram: https://t.me/s/gonjeshkedarandeofficial

This report from 2024-02-21 states: "The escalation comes at an inopportune time for Iran, which is also among world leaders in terms of using cyber warfare as a tool of statecraft. We assess Iran will diminish its overt military footprint and focus more on activity in cyberspace." My thinking is that if Iran will, in fact, be more active with cyberattacks, it makes high-profile retaliatory cyber-attacks such as those affecting Iranian critical infrastructure more likely.


Files
New Prediction

I am going slightly up because of the tensions around Taiwan (good weekly updates are published by ISW - recent one is here) and the already present political pressure on the Microsoft related to their business in China (you can read about it for example here). If something big happens around Taiwan (or in some other place because of hostile actions of China) US law-makers could force Microsoft to leave China or the pressure on them might be so big that they would do that by their own decision. I am not high on that, but this is something which I see as something which could lead to the resolution of this question as "yes".

Files
New Prediction
Why do you think you're right?

It is extremely unlikely now that he will cease to be the president in one month's time. 

Files
Why might you be wrong?

If that were to happen, it would have to be because of a High-Impact, (very) Low-Probability Event - something like a deadly accident or assassination attempt. Recent terrorist attack at Crocus City Hall, which was conducted most likely by ISIS might make us think that it is now more likely they will go after Putin, but I think that if such an assassination attempt was planned by ISIS to be executed within a month, then they would not attack the Crocus City Hall first, because it would reduce any chances of such assassination plan success very substantially. And even if they would plan it, their chances are even smaller now when not only security services and police are on alert and went after their members and sympathizers, but I bet the whole population is on alert and watching for possible terrorists.

Files
New Prediction

So, I looked deeper at the data, and as I suspected in my previous comment (and my wife knew it already) behind the outlier results in 2014 and 2016 was one specific mission: Kepler space telescope. Kepler was responsible for the discovery of 90% of exoplanets discovered in 2014 and 85% of the discoveries made in the record year 2016. It was also responsible for 49,9% of discoveries of exoplanets in the dataset.


What is important, Kepler was retired by NASA on October 30, 2018 after it ran out of fuel. In the dataset, we can still see discoveries of exoplanets attributed to Kepler after 2018 - even last year (70 discoveries, we can say it was a good year for Kepler), but the numbers are lower now (2021 was still the third-best year for Kepler) with an irregular downward trend. I assume that the discoveries were made in later years based on the analysis of images previously collected by Kepler.  

Its mission is continued by Transiting Exoplanet Survey Satellite (TESS) since 2018, but it was not that effective as Kepler, but it seems to be getting better over the years. See my table below with the most efficient projects/observatories over the years. Relative formatting is applied to each row to show changes over time.

Kepler might still be a source of a substantial number of discoveries of exoplanets this year (last year it was 70), or it may bring little to the total number (in 2022 it was 4), but we should not expect numbers like in 2014 or 2016 from it. Based on that, I am further reducing my probability slightly to 56%.

My thoughts: Base rates and Laplace rule are useful for forecasting, but if the causal mechanics/driving forces behind the numbers change, our expectations should also change. But this correction, of course, becomes subjective, and may make us even more vulnerable to the influence of biases. See a great comment about that by @ctsats : https://www.infer-pub.com/comments/121097


@ctsats @WeirdAwkward @DimaKlenchin @ScottEastman @probahilliby @MrLittleTexas @BlancaElenaGG @Tolga 




Files
ctsats
made a comment:

Thanks Michał. Just some secondary remarks.

Interacting with our spouses for relevant information is great 😀. But truth is, the stuff about Kepler's contributions were largely included in the "Discussion and analysis" part of my rationale below:

The relatively high numbers of yearly discoveries between 2013 and 2017 can be largely attributed to the Kepler space telescope, launched in 2009 and retired on Oct 2018, which had the detection of exoplanets as its primary mission. Although new exoplanets have been discovered from Kepler data as recently as 2023, its last contribution to the habitable worlds catalog were 2 exoplanets in 2020 (Kepler-1701 b and Kepler-1649 c).

In fact, the contributions of Kepler are even higher than your table implies, since 'K2' is just the name of Kepler's "extended" mission (2014-2018); in other words, sensor-wise, K2 is Kepler: https://science.nasa.gov/mission/kepler/ (K2 was not actually a time extension, just a renaming of the mission after Kepler lost a gyroscope in May 2013 and engineers managed to come up with a solution to save it).

Having this detail clarified reveals that the "long tail" of Kepler data is even more impressive and astonishing: anything labeled as "Kepler" in the files comes from data that were actually collected before May 2013 (the date when the 2nd gyro failed and data collection was disabled), including the high outlier numbers of 2014 & 2016, Kepler's "third-best year" (2021), and the 70 exoplanets confirmed in 2023! In fact, we had a data collection gap of more than 1 year, since K2 became operational only in June 2014.


Chronologically speaking, TESS went into operation the same year Kepler/K2 was retired (2018). The slight discrepancy between the number of exoplanets discovered by it in your table (415) and in my rationale below (420) is due to the 5 exoplanets confirmed by TESS in 2024 (ask for 2024 in 'Discovery Year" here). This gives us an indirect, and perhaps useful, hint: despite our resolution source having been reportedly last updated on Feb 1, it evidently does not include (yet) exoplanets discovered in January 2024., i.e the absence from the excel file of any discoveries in 2024 is due to an actual lack of update, and not to a lack of confirmed discoveries.

What your table nicely reveals is that, in a problem where we seem to have difficulty discerning clear and persisting trends, TESS rate of discovery seems to be steadily increasing indeed! But again, this is only about the total number, and not the habitables one (will look into it further, but TESS was responsible for the sole habitable detection of 2021...)


2021 was still the third-best year for Kepler

This in fact illustrates nicely what I and @Tolga have already observed about the practically non-existent correlation between the numbers of total exoplanets and habitable ones: despite being the 3rd-highest year for Kepler, 2021 has been so far the worst year for habitable worlds detected: only 1 planet (not even by Kepler, but by TESS), which can easily be considered a low-range outlier.


we should not expect numbers like in 2014 or 2016 from [Kepler]

We should definitely not, but again, speaking about the quantity of interest here (habitable worlds), we don't need the high numbers that were registered in those years (8 and 11 respectively) - a mere number of 5 will be enough to resolve this as Yes. And even if we subtract the last contributions of Kepler/K2 (2 habitable worlds among the 10 registered in 2020), we are still well above the required threshold for that year. We reached the threshold of 5 in 2022 and 2023 without any contribution from Kepler/K2.


Base rates and Laplace rule are useful for forecasting, but if the causal mechanics/driving forces behind the numbers change, our expectations should also change.

You are of course very right! Thing is, gaining the necessary familiarity with the specifics and the significant details of a forecasting problem in order to (hopefully begin to) understand such causal mechanisms and driving forces takes time (and collaboration!), and it can seldom (if ever) be assumed from the 1st (or 2nd...) sitting. That's why the base rates are only a start, and never the end, of the forecasting process.


PS And yes, just as I was ready to press "submit", the clarification I had requested was published, moving the resolution date to Jan 31, 2025 (or even later): https://www.infer-pub.com/questions/1373-will-at-least-five-more-exoplanets-be-found-to-be-potentially-habitable-between-1-february-2024-and-31-december-2024/clarifications


@ScottEastman 

The James Webb telescope has discovered 2 exoplanets

These are actually still candidates; as @jrl says, there is a standard procedure for an exoplanet to go from candidate to confirmed status. Fact is, as @HarrisonD has reported below (and I have confirmed myself), the JWST has so far contributed one single confirmed exoplanet in May 2023, and it is not a habitable one (neither are the new candidates, it would seem).


back to @michal_dubrawski

Do you by chance know how many candidates are still there among data collected by Kepler?

It seems we can get the official numbers. From the links here, go to 'KOI Table" (either entry, they both give the same result) and put CANDIDATE in the field 'Exoplanet Archive Disposition'; it gives 1983 records. Notice that this number is about Kepler proper (i.e. data before May 2013), and it does not include K2 (data from June 2014 to October 2018). For K2, go to K2 Planets and Candidates Table from the same page, and ask again for CANDIDATE in the 'Archive Disposition' field; it gives 1370 records. So, it would seem that we still have 3353 candidate exoplanets waiting to be confirmed (or not) from the Kepler/K2 data alone; if we accept a rough estimation of ~5000 current candidates (have not managed to get the exact number from the NASA archive, as asking for Candidate in the Solution Type includes candidates that have been subsequently confirmed), it means that, as of Feb 2024, ~67% of the candidates are still due to Kepler/K2.


@jrl

JWST's primary mission is cosmology, so they're not actively seeking exoplanets, and any that are discovered are probably just by chance.

This was my initial impression, too. But as already commented below, the study of exoplanets is included explicitly in its main mission objectives. What this means for discovery of new ones is less clear, but again, this is not of relevance for the question here, as it seems very reasonable that the further study of confirmed exoplanets can provide data for inclusion to the habitable list, which, after all, is what we actually care about here...

Files
Files
Tip: Mention someone by typing @username