Blog post: deepmind.google/discover/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/
Whitepaper: storage.googleapis.com/deepmind-media/DeepMind.com/Blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/AlphaEvolve.pdf
Basically they require users to hand-code a metric and provide a program skeleton with some parts of the code marked to be replaced, and then the system focuses on modifying the code regions in question to optimize the metric.
All the novel results they announced were in constraint satisfaction problems or optimization problem. Their results are still awesome, but it's not very different from AlphaGo style things.
Multiplication of matrices of specific size by 
Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-21
Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-21DeepMind likes coming up with new improved algorithms for these more specific cases, e.g. it was announced in 2025 that AlphaEvolve found a novel 4x4 complex valued algorithm that uses 48 multiplications.
Bibliography:
- fmm.univ-lille.fr/ attempts to keep an up-to-date list for various sizes
Steve Jobs' Apple-1 sells for $945k
. Source. Semiconductor device fabrication bibliography by Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-21
Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-21
Saint Eligius by Petrus Christus
. Source. Eligius pool is named after Saint Eligius, patron of goldsmiths and miners[ref]
Magic: The Gathering is Turing complete by Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-21
Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-21Related: Antihydra in Magic: The Gathering.
Updates Translation of Xi Jinping saying those against raise their hands by Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-21
Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-21 Updates Understanding the state of 3x3 matrix multiplication by Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-21
Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-21After yet another awesome announcement by DeepMind that it had improved theoretical 4x4 matrix multiplication reducing the number of scalar multiplications with its AlphaEvolve system, I decided to have a look at the smallest open size 3x3 to understand what was going on in there.
Updates Two Linux Kernel Module Cheat videos by Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-13
Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-13I made not one but two quick presentation videos about my project Linux Kernel Module Cheat, an emulation setup to study and develop the Linux kernel and more:
- www.youtube.com/watch?v=HDJFyCma32U: a presentation of me talking about it, edited up from my earlier presentation at Aratu Week 2024
- www.youtube.com/watch?v=fgDhe1tN50o: a demo of me running actually the project
I had meant to do this editing for a while and kept pushing it off because editing hurts, but finally sat down did it, partly prompted by my quick recent updates made to the projects part of post OurBigBook job search round 2025. At first I was thinking of making a single video, but after I recorded the demo a bit it seemed like two separate ones would make more sense.
I also created a bug report for Kdenlive, the video editor that I used, for a freeze that happens if you try to shift + delete the last item of the timeline: bugs.kde.org/show_bug.cgi?id=504103. Kdenlive is a good editor, but unfortunately it has new freezes and crashes relatively often.
One more useless task that I get off my head, on to the next!
Linux Kernel Module Cheat presentation
. Source. Linux Kernel Module Cheat demo
. Source. CIA 2010 covert communication websites HTML title element by Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-07
Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-07The discoverty of a possible HTML information leaks on HTML motivated us to download all HTML and have a grep.
<title> of webofcheer.com which is cryptically set as:pg1cWe started grepping with:and to just get the titles alone for visual inspection:
grep -ai '<title>' */index.htmlgrep -ahi '<title>' */index.html | sed -r 's/^\s*<title>//;s/<\/title>.*//'Some mildly interesting facts include:It is impossible to tell if these were oversights, or intentional to simulate common web development quircks. But they are cute in any case.
- opensourcenewstoday.com is titled just as "Title"
opensourcenewstoday.com/index.html:<title>Title</title> - a few sites are titled "Untitled Document" e.g.:This may have been the default title in Adobe Dreamweaver.
media-coverage-now.com/index.html:<title>Untitled Document</title> newsandsportscentral.com/index.html: <title>Untitled Document</title> newsincirculation.com/index.html:<title>Untitled Document</title> newsworldsite.com/index.html:<title>Untitled Document</title> primetimemovies.net/index.html:<title>Untitled Document</title> unganadormundial.com/index.html:<title>Untitled Document</title> - some others have empty title:
aeronet-news.com/index.html:<title></title> al-rashidrealestate.com/index.html: <title></title> arabicnewsunfiltered.com/index.html:<title></title> dailynewsandsports.com/index.html:<title></title> electronictechreviews.com/index.html:<title></title> indirectfreekick.com/index.html:<title></title> iran-newslink-today.com/index.html:<title></title> iraniangoals.com/index.html:<title></title> kickitnews.com/index.html:<title></title> mediocampodefutbol.com/index.html:<title></title> middle-east-newstoday.com/index.html: <title></title> mygadgettech.com/index.html:<title></title> sayaara-auto.com/index.html:<title></title> techwatchtoday.com/index.html:<title></title> the-open-book-online.com/index.html:<title></title> thenewsofpakistan.com/index.html:<title></title> theworld-news.net/index.html:<title></title> todaysengineering.com/index.html:<title></title> todaysnewsreports.net/index.html:<title></title> worldnewsandent.com/index.html:<title></title> - some others are titled just "index" or a variant of it:
all-sport-headlines.com/index.html:<title>index</title> europeannewsflash.com/index.html:<title>Index</title> fgnl.net/index.html:<title>Index Page</title> iraniangoalkicks.com/index.html:<title>index</title> just-the-news.com/index.html:<title>index</title> mide-news.com/index.html:<title>index</title> mytravelopian.com/index.html:<title>Index</title> noticiasdelmundolatino.com/index.html:<title>index</title> pakcricketgrd.com/index.html: <title>index</title> pangawana.com/index.html:<title>index</title> sportsnewsfinder.com/index.html:<title>index</title> thenewseditor.com/index.html:<title>index</title> turkishnewslinks.com/index.html:<title>index2</title> wahidfutbol.com/index.html:<title>index</title> webscooper.com/index.html:<title>index</title> webworldsports.com/index.html:<title>index</title> - a few don't have
<title>at all:b2bworldglobal.com/index.html bailandstump.com/index.html businessexchangetoday.com/index.html commercialspacedesign.com/index.html court-masters.com/index.html flyingtimeline.com/index.html marketflows.net/index.html nouvellesetdesrapports.com/index.html senderosdemontana.com/index.html sixty2media.com/index.htm
CIA 2010 covert communication websites Binary files by Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-07
Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-07As per:a few of the HTMLs are interpreted by grep as being binary:
grep . */index.html | grep 'binary file matches'grep: china-destinations.org/index.html: binary file matches
grep: classicalmusicboxonline.com/index.html: binary file matches
grep: driversinternationalgolf.com/index.html: binary file matches
grep: familyhealthonline.net/index.html: binary file matches
grep: grubbersworldrugbynews.com/index.html: binary file matches
grep: hai-pow.com/index.html: binary file matches
grep: hi-tech-today.com/index.html: binary file matches
grep: networkofnews.com/index.html: binary file matches
grep: nigeriastar.net/index.html: binary file matches
grep: noticias-caracas.com/index.html: binary file matches
grep: theentertainbiz.com/index.html: binary file matches
grep: thefilmcentre.com/index.html: binary file matches
grep: theinternationalgoal.com/index.html: binary file matches
grep: wildbirds-seasia.com/index.html: binary file matches
grep: worldedgenews.com/index.html: binary file matches CIA 2010 covert communication websites HTML analysis by Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-07
Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-07The HTML from the index page of Wayback Machine were:
- dumped at: github.com/cirosantilli/media/tree/master/cia-2010-covert-communication-websites/html
- downloaded with: github.com/cirosantilli/media/tree/master/cia-2010-covert-communication-websites/download-html.sh. Note that there were many supurious errors notably:we just ran it multiple times until all errors were gone.
OpenSSL SSL_connect: SSL_ERROR_SYSCALL in connection to web.archive.org:443
The best way to analyse the HTML is to grap our dumps from: github.com/cirosantilli/cia-2010-websites-dump.
Some possibly interesting searches include:
Some of the HTML files contain conditional comments e.g. web.archive.org/web/20091023041107/http://aquaswimming.com/ contains:
<!--[if IE 6]> <link href="swimstyleie6.css" rel="stylesheet" type="text/css"> <![endif]-->Varios of the non-English websites seem to have comments translating the content e.g.:This feels like it could be the translation helping the technical webdev team know what is what.
./noticiasmusica.net/20101230165001/index.html:<h2>Alguns dos Melhores Sites Nacionais</h2><!--some of the best national sites (in music)-->Many of the RSS frame pages use:which is a weird HTML tag that would lead all links to open on new tabs, e.g. web.archive.org/web/20110202124411/http://thecricketfan.com/home.html.
<base target="_blank" />Various websites have pages with .php extension. It feels likely that all websites were written in PHP.
Some sites use a
feeds.php for the feeds, e.g. http://www.absolutebearing.net//absolutebearing_feeds/feeds.php?src=http%3A%2F%2Ffeeds2.feedburner.com%2FOceanyachtsinfo&desc=1Some URLs existed both in HTML and .php extension, or were converted at some point:
allworldstatistics.com/20110207151941/comprehensivesources.html
allworldstatistics.com/20130818155225/comprehensivesources.phpA few of the PHP urls have weird IDs in them like we wonder what they mean.
omktf, juqwt and qlaqft:./middle-east-newstoday.com/20100829004127/omktf/uirl.php?ok=461128
./newsandsportscentral.com/20100327130237/juqwt/eubcek.php?pe=747155
./pondernews.net/20100826031745/lldwg/qlaqft.php?fc=281298A few separate websites have an archive with the same It is unclear what it means. All of them contain something like:so looks like an archival artifact only.
pid parameter:fightwithoutrules.com/20131220205811/?pid=2POQ7BC1G/index.html
half-court.net/20131223165013/?pid=2POQ7BC1G/index.html
health-men-today.com/20131223002237/?pid=2POQ7BC1G/index.html
intlnewsdaily.com/20131221121441/?pid=2POQ7BC1G/index.html
intoworldnews.com/20131217193621/?pid=2POQ7BC1G/index.html<html>
<head>
<meta name="robots" content="noarchive" />
<meta name="googlebot" content="nosnippet" />
</head>
<body>
<div align=center>
<h3>Error. Page cannot be displayed. Please contact your service provider for more details. (11)</h3>
</div>
</body>
</html>The following two websites have a
feeds.php system for their RSS:./mydailynewsreport.com/20110211111053/myrss/feeds.php?src=http:/www.refahemelli.com/pashto/news/rss.php&chan=y&desc=1&targ=y&utf=y
./magneticfieldnews.com/20110208063545/magneticfeeds/feeds.php?src=http:/www.bbc.co.uk/pashto/index.xml&chan=y&desc=1&targ=y&utf=ySome of the HTML uses attributes without quotes, which is legal, but very unusual nowadays:
soldiersofsouthasia.com/20110207203705/home.htm: <a href=http://www.rss-to-javascript.comWe can try to search for any link leaks by listing all domains linked to with:The first thing that shows up is that there are some IPs linked to directly! But they seem to be the direct IPs of legitimate websites, we are not sure why IPs were used rather than domain names:
git grep --no-color -I -h --no-line -o 'https?://[^/">?]+[/">?]' | sed -r 's/.$//' | sort | uniq -c | sort -nk1- 69.167.160.171 at web.archive.org/web/20110208053653/http://sa-michigan.com/ to web.archive.org/web/20100304122019/http://69.167.160.171/ marked with image "fantasyplayers.com", a legit website called Fantasy Players Network
- 69.94.11.53 at web.archive.org/web/20101229193800/http://newsresolution.net/ titled "International Tribunal for Rwanda" to web.archive.org/web/20101229193800/http://69.94.11.53/default.htm
- 74.125.77.132 mynepalnews.com Webalizer
- 194.165.154.66/index.php web.archive.org/web/20110129161937/http://icwb-news.com/ MiddleEast links to 194.165.154.66/index.php but that is an actual page: web.archive.org/web/20110529142501/http://194.165.154.66/index.php
- 200.55.6.87 at web.archive.org/web/20110128170204/http://noticiasdelmundolatino.com/ after clicking "Maps" tab entitled "Mapas en la red" to web.archive.org/web/20100329150648/http://200.55.6.87/es/index.htm
- 213.97.154.118 at web.archive.org/web/20120429042725/http://montanismoaventura.com/ entitled "Mallorca Verde" to web.archive.org/web/20120430191214/http://213.97.154.118/mallorcaverde/ The target is a bit weird and almost empty.
- 216.218.196.146 at entitled "AskTheDr.com" to web.archive.org/web/20070303080403/http://216.218.196.146/askthedr/index.htm
We can also get the full line for each with sorted by least common domains with the slow:
git grep --no-color -I -h --no-line -o 'https?://[^/">?]+[/">?]' | sed -r 's/.$//' | sort | uniq -c | sort -nk1 | awk '{if ($1 < 10) print $2}' | xargs -I{} git --no-pager grep -h --no-line -o '{}.*<' | tee tmp.log
We can search for all IP-like strings with:
git grep '[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\b' Linux Device Drivers book 3rd edition by Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-07
Ciro Santilli 37 Updated 2025-07-01 +Created 2025-05-07lwn.net/Kernel/LDD3/ contains the PDF of each chapter.
There are unlisted articles, also show them or only show them.