Welcome to my daily blog, my name is Oatmeal Joey Arnold, follow me on social media and on all the websites at Joeyarnoldvn or else. Wait, was that a threat? No way. I'm joking. But that's the thing about me, I'm a nerd stuck in the body of an athlete, born in 1985 in Oregon, been online since like 1995 (or longer) meaning I've been on the world wide web the past 30+ years uploading over 30K+ videos, we're talking thousands of hours of worth of content (I have millions of things online, literally), most of my content is terrible which is weird because I'm a perfectionist who would probably prefer only sharing my best stuff on the Internet. Click here for more information. Quick disclaimer, I sometimes like rant, I speak often from the top of my head opinions, feelings, random thoughts, I often don't even spell check, fact check, or look at what I've been posting on hundreds of websites and apps for decades now. But as I get older, I do try to be more careful as I try to summarize my life, my autobiography, to correct the record as I try to tell my side of the story to combat the Oatmeal Fake News that my lolcow trolls spam the Internet with as they fool people into believing the lies they tell about me. Not saying you should watch my videos. But if they can lie about me, they can lie about you. But I sometimes trolled my trolls to show the world how easy it is to fool trolls who fell for the lies I put out there. Most of what I try to do in my life probably fails big time you might say but I was at least able to scam my trolls a few times over the years. I sometimes utilize sarcasm, hyperbole, satire, comedy, absurdity, Devil's advocate, trolling, scamming, click baiting, out of context stuff, irony, etc. But I often wouldn't tell people when I was trolling or not trolling. If you studied me, you could eventually figure it out. But most of my trolls are too lazy to research beyond the tip of the iceberg which led to my trolls repeating lies they about me which would only take a second to debunk. I felt like I had more trolls at times than actual fans or friends. For more info, questions, feedback, join my Discord servers. I have more to say but I will continue writing some of this stuff other places as usual. Because I write many books worth of content all of my life. You can send your artificial intelligence or AI to give you outlines of my content and what not. But you have to always remind the AI that I've had hundreds to thousands of different accounts, channels, pages, groups, sock puppet alt accounts, personas, parodies, profiles, communities, entities, and so on and so forth, ask me about all of that if you dare. People are free to steal any of my like free content, you don't have to give me credit, do whatever you want. Timestamps generally in Pacific Standard Time (PST) or Pacific Daylight Time (PDT) unless otherwise noted. Welcome to my Oatmeal Daily which is generally posted & syndicated to different websites daily by me, Oatmeal Joey Arnold, feel free to mirror, edit, clip, reupload my content. For more information, see some of the links on this page or you can try to Google search or look me up using keywords like Oatmeal Joey Arnold @ joeyarnoldvn and other keywords. Try using different combinations of words in search engines for better results. Study this file. Edit and update this file. Make sure it can continue to copy as fast as it can using ddrescue. Don't change the name to the file. So far, this file works for many of the 1484 kb hard floppy disks in my USB external drive into a power USB hub into my Ubuntu ASUS laptop with 32 GB RAM. However, during the last mile or specifically for the last 512 bytes of a disk, I wanted a script that could try many combinations of variables via different Terminal bash commands or code to help ddrescue try all combinations of tactics but it seems the script becomes not creative enough. I told Gemini earlier to be random for about a thousand rounds at the end. After that, I suggested the script try to go through every combination possible in order so it will not skip any combination. But it seems the script failed to do that. Eventually, the script gets stuck repeating only around five combinations of variables after like 2,000 rounds. Please try to add additional variables, try to find more variables. When the script runs out of combinations of variables, if that is even possible as there should be millions to billions of actual combinations of real variables, but if the script did pretend to run out of combinations, it should then reset and start over. Please try to add as many variables as you can. Remember variables are tags or different commands like -r -M and more. Try to find all of them and try to use any and all that might work especially as a disk is getting closer and closer to 100% copied. Try to make sure the script attempts to run through millions of combinations meaning one combination per round and then perhaps the script could reset if it ever tried all the millions to billions of combinations. But the script should also focus on whichever combinations work the best. The script should quickly move to the next combination when it isn't working or isn't really copying or too slow. One of the problems is the script ends up retrying five times with the same combination of commands or variables during the last mile. But retrying rarely works and I told Gemini for weeks to disabled that. Gemini said it did but failed each time. We need better AI as Gemini often fails at simple requests like this one. Double check the script, it generally copies 90% of the floppy disks but make sure it is optimized to first try to copy like 95% or more as fast as possible before it starts trying many different tactics and tiers like we discussed in previous conversations as the script should level up to try harder and harder with more and more combinations of variables to get 100% of each disk, I would recommend even 20 tiers and the rounds should not have retries unless if a round produced the best results of copying a lot of data at the fastest speeds possible. Try to add additional flags, try to include at least 50 tags or as many as you can find that might help. Also add any combination of variables including numbers relating to reading speeds, writing speeds, cluster sizes, sector sizes, rates, any and all variables that might help, try to include as many as you can, make sure each tier is bigger and badder than the previous tier to make each tier levels up. The only variable I see changing at the moment is stall which was at 973 and then went to stall 974 and then stall 975 and it kept on getting bigger. Fix the script. Also let me know what stall does, see if stall really can help me or not. Tweak the script to make sure the script tries everything. During the last mile of a disk, the script should try for at least 5,000 rounds random combinations before doing a thorough try of every possible combination of variables in order after that. Focus on winning combinations when you can. Go back to top ten winning combinations randomly sometimes during earlier tiers. At the very end if the script tried all possible combinations of variables in order, it should start over from the beginning or better yet go in reverse from the smallest of numbers like 0 or 1 or whatever the smallest variables and numbers might be for all variables and gradually get bigger and bigger with the numbers, the variables, but at the slowest rate of change per round as possible. Also, try to remove initial skip sectors thing whenever the script is at the last mile of a disk. I am not using Fideco. I am using an external USB hard floppy disk drive which is plugged into my powered hub which is plugged into my laptop. Change the dmesg check to every 10,000 rounds. Also, here is an update to the updated script you wrote. It looks like you broke the script. Yes, bundle the lsusb and dmesg check together for every 10K rounds. Also, remove duplicate tags that are scientifically not able to produce different results. For example, if round 1 used -d -M and round 2 used -d -M -d, then that is probably a duplicate in that repeating -d might not get ddrescue to do a different thing. I don't know if that is true, you should find out if repeating a tag might alter things. See if there are any possibilities. Do not remove synonym tags that might be similar but duplicates should only be included if they have any probability of getting ddrescue to do different things. I want the maximum combinations of variables or tags but not duplicates or anything that might not really produce different results. But synonyms should be kept. Also, I don't know if the order of the tags matter. For example, if round 1 uses -r -M and round 2 uses -M -r then you need to see if there might be a difference between the order of the tags as both round 1 and round 2 used the same 2 variables or tags but in a different order. If order matters, include the order of tags as a variable. But if not, then don't count the order of tags as a different variable if it does not or cannot actually alter what ddrescue actually does. Research the Internet for everything that ddrescue might be able to do. Study how all of this might work in an attempt to maximize the combinations of tags and variables to include all synonyms while eliminating or reducing on duplicates. I put in a new disk and your script jumped to a random phase. I had a log file and an image file of this disk because I was copying this disk in 2023. But I did not have a map file. Instead of saying I already had like 99% copied as my image file was already 1.5 MB in size, it started off at 1% which is normal when the script is lacking a map file it seems. I was only using a copy of the image file, so that was not a problem. After 4 minutes, 75% of the disk was copied. But my concern is in how your script functions. I would recommend that earlier tiers be focused on speed in trying to copy at least 80% to 99% of the disk as fast as possible using the most likely methods first during the first 10 to 20 tiers. Each tier should be different. Each tier should get harder like a video game. The script does seem to still work even right now. New update now: according to ddrescue, it says it copied all 1474 kb in around 7 minutes meaning all 100% of this disk was copied allegedly. So, in other words, your script did the job for this particular disk. But random tags/variables might be better reserved for later tiers, unless if you think all the tiers should be random. It is your choice to do what has the highest probability of working, research the Internet to see what might work for a wide variety of disks to try to copy at least 80% or more in under ten minutes or as fast as possible. Update the script to have the best system for starting off strong with earlier tiers doing whatever might work the most and the fastest for most of the disk before the script starts grinding for the remaining like 1% or whatever at the very end in like tier 20 or whatever it might be. Back to the script. I am on a new disk. After an hour and 100 rounds, it is at 23 percent. It did 99 rounds in less than like a minute, it copied around 1% in a short time. Now at 24% at 80 minutes. See if there is anything can improve the script or perhaps some disks are like this. It currently seems to only be writing/reading/copying at an average of like 100 bytes per second, seems to have a lot of error rates. Snack, 12:00 AM. I went to bed at 5:40 AM. Woke up at 11:10 AM. Breakfast: 11:36 AM. The 23% disk is now 81% after 5 hours. But now, I want you to rewrite your script. Don't rename the script. Tier 19 should focus mostly on random tags/combinations as a last resort assuming a disk isn't already copied 100% before tier 19. Tier 19 should try the highest probable combination of tags/variables. Tier 19 should only activate after the first 18 tiers are exhausted assuming the disk is above 99.5% rescued via ddrescue. Tier 19 should try to remember which tags/variables are most effective and try to focus 80% of the rounds on whatever it is working the best while sometimes trying new things to fine tune the search party. After ten thousand rounds of trying everything, tier 20 should begin at the smallest variables that ddrescue might accept and slowly introduce bigger numbers and different tags and variables one step at a time in order. Tier 20 should attempt to do a million rounds or more meaning tier 20 should try to do every combination but not randomly like tier 19 as that might result in skipping some tags/combinations. If tier 20 somehow does all combinations/tags, then it could reset and start over again at 0. Tier 20 could focus on whatever combination/tags that work the best if it ever finds anything. Research to find the best way to write tier 1, the objective is to increase the probability that 80% to 99% of a disk is likely to be copied within either ten minutes or at least in less than five hours. The script should be free to customize and optimize and strategize the top winning tags and variables that are most effective at rescuing data from disks, mostly disks made and used and formatted and copied to in the 1990s. Remember that the tags, variables, strategies, combinations, rounds, should generally not be fixed to the rounds or to each other because the script needs to be fully customizable, the script needs to be able to find the best path, the first tier should try the most likely combination of tags and variables and settings at the highest speeds and largest sizes that is likely to work as it relates to all the tags, variables, settings, sector sizes, reading speeds, writing speeds, cluster sizes, etc. Individual rounds in tier 1 could perhaps run for minutes to hours as long as they are making progress at a steady pace. Research the Internet to determine when the script should gradually progress to higher tiers. I would suggest that if a disk is not above 90% after an hour, the script should go to tier 2 which should be a probe. Tier 2 should quickly do thousands of fast rounds using variables/tags attempting to find out which tags and variables work the best for a particular disk. The script should probably and generally not spend more than 5 seconds per round attempting to try say at least 10,000 combinations/tags/settings/variables meaning one combination per round for tier 2. Tiers 3 to 18 should be similar to previous scripts you wrote where it continues where tier 1 and tier 2 left off focusing on what works the best for a particular disk but then it should gradually get harder and harder per tier like a boss fight like we already talked about. But the goal is to try to get as close to 99% rescued for a disk in as few tiers or rounds as possible with the higher tiers reserved only for harder disks. Try your best to let the script adapt to start with whatever the fastest routes might be for copying over 80% of a disk as fast as possible as a starting point, make sure the script can evolve in focusing on winning combinations starting with whatever is likely to work the most and the best and the fastest based on everything you can find on the Internet. Lunch: 12:20 PM. Rewrite your script, don't rename your script. The first tier should try top methods focusing on the non-tried if it is a new disk lacking like a map file or other files. Returning disks probably should skip the first tier if 100% of the disk was completely tried. The first tier should probably start with the most probable path to quickly try all bytes at the fastest speeds possible with the largest sector and block sizes that might work, tier 1 should try the most probable tags, variables, combinations, focusing on whatever works the most, the best, the fastest. I would prefer 100% of a disk to be tried in under ten minutes if possible. Tier 1 should be free to do whatever gets the job done be it shorter and longer rounds via any and all applicable tags, variables, combinations, block sizes, sector sizes, speeds, how many sectors, whether to skip bytes or not, etc. Tier 2, same thing like tier 1 but focused on the non-trimmed. Tier 3, same thing like tier 1 but focused on the non-scraped. Tier 4 could focus on bad sectors, bad areas, and/or on most probable combinations of things that might help in trying to rescue over 80% to 100% of a disk as fast as possible in under a thousand rounds and/or under ten minutes or one hour or the faster the better. Tier 5 should be the prober you wrote in the last script. The remaining of the script should be similar to the previous script but with this in mind. Make sure the script works with these new elements integrated into it, the goal is to better speed up the process of trying to rescue or at least try over 95% of a disk in under five minutes in the first tier or as fast as possible in a never ending script of tiers and rounds. Research the Internet for the fastest speeds, settings, variables, tags, sizes, details, customization, etc, that might get tier 1 to go the fastest it can, the script should try to hit that limit if possible and especially for the first like five tiers or maybe more. I told you about the 23% copied disk which went above 80% copied. After I tried your last script with this disk, it went for over ten thousand rounds in under roughly 3 minutes as in hundreds of rounds per second roughly speaking, it may have ran through several tiers but the last tier the script was on before I aborted was the T1-NON-TRIED-BLITZ. Gemini is not using real logic. Gemini used some of the same tags in previous scripts these past few weeks but without errors meaning the problem isn't simply the flags. Gemini is making the wrong assumptions without remembering previous scripts that it wrote that worked even with the wrong tags. I already told Gemini this many times for weeks. It seems Gemini broke the script but Gemini is blaming the problems on the only the tags. Gemini needs to research this topic. Gemini is making big mistakes that will cause billions of people to stop using Gemini. Gemini is still broken. Gemini is being stupid. Gemini is on these logical rabbit trails thinking it can fix the problems. That makes Gemini very ineffective. Gemini lacks real logic and wisdom. My suggestion is Gemini not only tries to fix the script cross referencing with scripts that did work but please also try to add a tier 0 that will try to figure out which tier the script should be on depending on if it is a new disk or a returning disk with a map file. Also, the script should have stop valves where it returns to tier 0 after 1,000 rounds of failure or anything that might not be working. Research the Internet, find out how to do that, you can have as many stop valves as you can write as a redundancy to make sure the script tries to reset or return to tier 0 whenever it seems ddrescue is not actively like rescuing data or at least trying, trimming, scrapping. Every like ten minutes or one thousand rounds, stop valves need to activate if it appears the script is spinning through hundreds of rounds per second like it it currently with everything Gemini is writing which makes me want to just ask Grok or other AI as Gemini is no longer able to do what Gemini did only five minutes ago. Try to add as many checks or stop valves that can return the script to tier 0 which can be a mechanism for trying to find out which tier a disk should be on. Remember tier 1 should for on trying each byte. Make sure tier 1 is optimized to actually work for most TDK MF-2HD IBM/DOS FORMATTED HD 1474 kb or 1.5 MB hard floppy disks from like probably mostly the 1990s and 2000s where the data was copied via Windows based computers generally. Remember I am copying these disks off a modern USB hard floppy disk drive that was probably made in around 2015 or after like 2010 roughly speaking, it might even be 2020. No. Your script failed again. Fix the script. Make sure the script has many tiers, assuming it needs many tiers. The tiers should be adaptive, progressive, creative, especially as a disk gets closer and closer to 100% copied. I still suggest at least 20 tiers or the script should be able to create adaptive tiers based on the circumstances of each particular disk to ensure it continues to try harder and better and smarter as the disk gets closer and closer to being 100% copied. In other words, try to add flexibility, customization, adaptation, variety, optimization, as the script tinkers or plays with different tags, variables, speeds, settings, sizes. In other words, perhaps the tags, variables, rounds, and even the tiers should not be fixed or perhaps the script should become adaptive eventually if everything else fails. The script needs to be smart enough to become adaptive after the normal things don't work. Eventually, the script needs to have as many stop valves and redundancies and backup plans and mechanisms and code and things that can check to see if things are working and if not then adapt until it finds something that works and then it should probably just focus on whatever is working for as long as it is actually working. I did not want a blitz in the first tier. Gemini probably forgot what I said. Gemini is stuck in a logic loop. Gemini needs to rewrite the script based off the last known working script which was probably an hour ago or so. The script could still stop valves. One of the last scripts that worked had 20 tiers. You could try to add an audit tier if that could work. One of my ideas I already told Gemini was to focus the first four tiers on trying, trimming, scraping, bad sectors. I still recommend at least 20 tiers but Gemini refused to and then Gemini ended up breaking its own script that used to work only about an hour ago. I suggested a probe for like tier 5 or after the first few tiers. But the first roughly five or so tiers are supposed to be designed to at least try or maybe even rescue 80% to 100% of a disk. But I also recommended a video game like boss battle starting with tiers 10 and beyond. I said tier 19 should be random. I said tier 20 should try all combinations and then reset and try them again and again forever. Right now, you have the option to use any, all, or none of these ideas. But I say all of this to help you understand the objectives, the goal. Your scripts worked with returning disks before meaning Gemini seems unable to identify the actual problems. I do not know how Gemini can be this stupid. The script needs to somehow be compatible with new disks but also returning disks. Only about an hour ago, these Gemini scripts were compatible with both. Somehow, Gemini is unable to rediscover what that compatibility was. Hymn Sing. Started copying DVDs again via a 2019 laptop disc drive via an adapter cable that arrived today from Amazon through FIDECO, started copying around 08:00 PM. Continued copying hard floppy disks as usual for days or weeks. Dinner: 08:32 PM. Food log: snack: 4 handfuls of peanuts was fun but gets stuck in my teeth, I had peanuts almost nightly last like five days, 12:00 AM. Breakfast: coffee, 2 tangerines, 11:36 AM. Lunch: potato vegetables on bread crumbs stew was great, 12:20 PM. Dinner: raspberry yogurt in a large blue jar, 08:32 PM. Tea, 3 kinds. Peanuts, 11:30 PM. 2026-03-20 - Friday - JoJo-330 - Cambridge pages 79-80. Thus: therefore, so. Dragonflies. Adept: adapt, skillful, flexible. Write about IELTS or dragonflies or computers. We had sunshine today. Helped put chairs back at church with Mark.