Category Archives: Isaac Caret

In the last few days, looking at the whole Isaac / Chad ‘alien alphabet’ mystery has made me think more broadly about cipher mysteries. What I’m trying to do is to work out what the relationship between the different pieces of evidence are – but not just in terms of “A preceded B”.

Inserts

Generally, the practical problem with cipher mysteries is that the relationship between “layers” isn’t just ‘archaeological’, i.e. they aren’t just laid down one on top of the other. Very often we find ourselves looking at annoying evidence where one layer pretends (or, more charitably, ends up appearing) to be out of order. The term I typically use for this is “insert” (but please let me know if there’s a better word or phrase!), to denote something that someone has attempted to insert into the timeline.

In the case of Isaac / Chad, I can’t help but wonder if Isaac saw the strange diagrammatic detailing on the large cropped image released by Ty B and built his entire account out backwards from there, to try to insert his own (fake) account into the pre-drone-sighting timeline?

Remember, Isaac wrote (having disclaimed any connection with the drone observation people):

More importantly though, I’m very familiar with the “language” on their undersides seen clearly in photos by Chad and Rajman, and in another form in the Big Basin photos.

Yet Isaac also wrote:

It’s no surprise that these sightings are all taking place in California, and especially the Saratoga/South Bay area. Not far from Saratoga is Mountain View/Sunnyvale, home to Moffett Field and the NASA Ames Research center.

As far as I know, Saratoga was only properly identified as the location for some of the photos a long time after Isaac wrote this, so this section does conversely suggest that there was cooperation / coordination. It’s hard to read how these things all fit together.

Missing evidence, Google problems?

The presence of the higher-resolution dragonfly drone image in the Project Avalon set suggests to me that I’m in fact dealing with scaled-down versions of larger images, but where the EXIF data has been preserved across the scaling-down. And so I’m now hungry to find even earlier (and larger, and unscaled-down) versions of all these images.

However, I have to flag that I’m a bit concerned about Google. In the past, I’d be really confident that Google Images would find a whole load of images: but now it feels as though this whole part of Google’s search engine has been gamed by Pinterest and others (Japanese blogs seem to be good at this, oddly). Basically, I’m not even getting 10% of the results I used to get, and the quality of the results I do get has dropped right down too.

I’ve had similar experiences with Google’s main text search recently, where queries that I have previously used to find things now don’t work at all. Whereas I used to save query strings in my notes to help me find groups of related things, that strategy seems to be working less and less well over time. More generally, I’m finding it harder and harder to find things online, and for the kind of research I do, that feels like it is growing into a huge problem.

People may post endlessly about the death-spiral that Elmo’s Twitter has apparently entered, but I can’t help but wonder whether Google too is now entering some kind of mysterious end-of-life phase? Perhaps you’ve noticed this too.

When “Isaac” posted his alien alphabet / antigravity stuff in June 2007, it was (he claimed) in response to recent reports of strange ‘dragonfly’-shaped drones, some of which had the same ‘alien’ writing on them:

• 10 May 2007, Bakersfield, California – “Chad” (April 2007 + 06 May 2007)
• 12 May 2007, Lake Tahoe, Nevada – “Deborah McKinley” (05 May 2007)
• 20 May 2007, Capitola, California – “Raj / Rajman / Rajinder Satyanarayana” (16 May 2007)
• 06 Jun 2007, Big Basin, California – “Stephen” (05 Jun 2007)
• 11 Jun 2007, Big Basin, California – “Ty Branigan” (05 Jun 2007)

Those sightings are well documented in a “One Year Later” article in MUFON Ufo Journal April 2008 (pp. 3-11): the TL;DR version of that is simply that none of the claimed witnesses is credible, sorry. [If you don’t know about MUFON, it describes itself as an independent follow-on to Project Blue Book, and that it always starts by assessing the credibility of witnesses.]

Regardless, I decided (as I did with Isaac’s JPEGs) to take a digital forensic look at the various drone images, to see if there was anything interesting there: and I began with “Chad”.

Chad’s drone images

Chad’s drone story first appeared on Coast to Coast AM, and starts as follows:

Last month (April 2007), my wife and I were on a walk when we noticed a very large, very strange “craft” in the sky. My wife took a picture with her cell phone camera (second photo). A few days later a friend (and neighbor) lent me his camera and came with me to take photos of this “craft”. We found it and took a number of very clear photos. Picture #1 is taken from right below this thing and I must give my friend credit as I was not brave enough to get close enough to take this picture myself!

I started by downloading numerous variations of Chad’s images, but (viewed through a JPEGsnoop microscope) none of them seemed to me to be an original image. However, once I found Chad’s images from the Coast to Coast AM website itself, JPEGsnoop had an absolute field day.

The first image I looked at in depth was “Craft050607b.jpg”:

There’s an absolute riot of EXIF metadata going on here. For a start, we can see that the image creator used Adobe Photoshop Elements, which is sold as a cut-down version of Adobe Photoshop:

[Software                            ] = "Adobe Photoshop Elements 2.0"
[DateTime                            ] = "2007:05:06 17:20:08"

We can also see that it was saved out in Adobe Photoshop with quality setting 5:

  8BIM: [0x0406] Name="" Len=[0x0007] DefinedName="JPEG quality"
    Photoshop Save As Quality                          = 5 
    Photoshop Save Format                              = "Standard"
    Photoshop Save Progressive Scans                   = "3 Scans"

Similarly, the creator used Adobe Digital Negative (DNG) Converter, which is typically used to import raw (unprocessed) photo data from (generally high-end) digital cameras:

 SW :[Adobe DNG Converter      ]                                [                ]                  
 SW :[Adobe Photoshop          ]                                [Save As 05      ]                  

Oddly, though, there was a slice name that seems to imply that this was from a scanned image:

      Name of group of slices                            = "ScannedImage-2"
      Number of slices                                   = 1 

UUIDs and Melissa

Finally: also embedded in many of the file metadata is a version 1 format UUID. Historically amusingly, its strategy for constructing a universally unique id is by combining a device’s six-byte MAC address with the number of 100-nanosecond ‘ticks’ since midnight 15 October 1582 UTC, which was (as I’m sure you’ll all recall) the precise date when the Gregorian calendar was first adopted.

Decoding Chad’s UUIDs using an online UUID Decoder gives us:

• “Craft050607a.jpg” has UUID 013dfdd9-fc30-11db-b305-b8a28f50b702
  • MAC address b8:a2:8f:50:b7:02, generated on 2007-05-07 00:15:04.703125.7 UTC
• “Craft050607b.jpg” has UUID 013dfdd9-fc30-11db-b305-b8a28f50b702
  • MAC address b8:a2:8f:50:b7:02, generated on 2007-05-07 00:15:04.703125.9 UTC
• “Craft050607c.jpg” has UUID c822a9ca-fc30-11db-b305-b8a28f50b702
  • MAC address b8:a2:8f:50:b7:02, generated on 2007-05-07 00:20:38.390625.0 UTC.
• “Craft050607e.jpg” has no embedded UUID
• “Craft050607x1.jpg” has UUID 8e3de69f-fd14-11db-a9e5-8988b4fa457e
  • MAC address 89:88:b4:fa:45:7e, generated on 2007-05-08 03:31:06.515625.5 UTC
• “Craft050607x2.jpg” has UUID 8e3de6a2-fd14-11db-a9e5-8988b4fa457e
  • MAC address 89:88:b4:fa:45:7e, generated on 2007-05-08 03:31:06.515625.8 UTC
  • This didn’t seem to use Adobe DNG).
• “Craft050607x5.jpg” has UUID 74fb1a10-fd17-11db-a9e5-8988b4fa457e
  • MAC address 89:88:b4:fa:45:7e, generated on 2007-05-08 03:51:52.625000.0 UTC
  • This too didn’t seem to use Adobe DNG.

At first glance, it might seem we have identified Chad’s two PCs! Except… we haven’t: the OUIs (the top three bytes) of the two MAC addresses are not recognised, so it is almost certain that Adobe’s software picked a random MAC address at the start of each of the two sessions. This is almost certainly because of privacy concerns: what famously happened in 1999 was that the creator of the Melissa computer virus was tracked down via his MAC address embedded in a UUID embedded inside his virus. And so people started randomising the MAC address portion of UUIDs, which eventually led to UUID format v4 and v5.

Hence: unless “Chad” happened to write out any other JPEGs during those two specific sessions, I think we are unlikely to be able to use the MAC address portion of Adobe UUIDs generated in 2007 (and afterwards) to track down anything else linked to this person, alas.

Finally: note that each of these JPEG files contains two UUIDs formed from timestamps that differ by:

• Craft050607a – 200 nsec
• Craft050607b – 900 nsec
• Craft050607c – 333 msec
• Craft050607x1 – 200 nsec
• Craft050607x2 – 200 nsec
• Craft050607x5 – 2.95 sec – this seems to be because the base UUID is that of Craft050607x2 (it was apparently cropped from that image and saved three seconds later)

I don’t really know what this means, but I thought I’d include it anyway.

Lake Tahoe images

I also found copies of the Lake Tahoe images on a (very helpful) Avalon Library page. The two files were called “7013_submitter_file1__070505_02.jpg” and “7013_submitter_file2__070505_03.jpg” (which I presume are partly MUFON case file references).

7013_submitter_file1__070505_02.jpg included some interesting metadata. Overall, JPEGsnoop’s assessment was this:

SW :[Apple ImageIO.framework ] [075 (High) ]

The sensor was annotated as “MSM6500” (which I presume is the Qualcomm chip), and there’s Adobe Photoshop XMP metadata in there too. For once, the EXIF data is consistent with a digital camera:

[ExposureTime                        ] = 1/21 s
[ExifVersion                         ] = 02.20
[DateTimeOriginal                    ] = "2007:05:05 18:52:11"
[DateTimeDigitized                   ] = "2007:05:05 18:52:11"
[ComponentsConfiguration             ] = [Y Cb Cr .]
[Flash                               ] = Flash did not fire
[FlashPixVersion                     ] = 01.00
[ColorSpace                          ] = sRGB
[ExifImageWidth                      ] = 0x[00000200] / 512
[ExifImageHeight                     ] = 0x[00000180] / 384
[CustomRendered                      ] = Normal process
[ExposureMode                        ] = Auto exposure
[WhiteBalance                        ] = Auto white balance
[DigitalZoomRatio                    ] = 2/1
[SceneCaptureType                    ] = Standard
[SubjectDistanceRange                ] = 0

The second image was captured five seconds later (also without flash):

[DateTimeOriginal                    ] = "2007:05:05 18:52:16"

Stephen images

Here, the EXIF data (in “IMG_1060”) is definitive about the camera used (the Rebel XT is a good camera, I have one myself):

[Make                                ] = "Canon"
[Model                               ] = "Canon EOS DIGITAL REBEL XT"

The other camera EXIF data is very much what you’d expect for a daytime shot:

[ExposureTime                        ] = 1/4000 s
[FNumber                             ] = F5.6
[ExposureProgram                     ] = Aperture priority
[ISOSpeedRatings                     ] = 1600
[ExifVersion                         ] = 02.21
[DateTimeOriginal                    ] = "2007:06:05 13:12:49"
[DateTimeDigitized                   ] = "2007:06:05 13:12:49"
[ComponentsConfiguration             ] = [Y Cb Cr .]
[ShutterSpeedValue                   ] = 7694/643
[ApertureValue                       ] = 7163/1441
[ExposureBiasValue                   ] = 0.00 eV
[MeteringMode                        ] = Pattern
[Flash                               ] = Flash did not fire
[FocalLength                         ] = 50 mm
[FlashPixVersion                     ] = 01.00
[ColorSpace                          ] = sRGB
[ExifImageWidth                      ] = 0x[000009C0] / 2496
[ExifImageHeight                     ] = 0x[00000680] / 1664
[ExposureMode                        ] = Auto exposure
[WhiteBalance                        ] = Auto white balance
[SceneCaptureType                    ] = Standard

This too had an Adobe XAP block, but without any UUIDs.

IMG_1061 was taken two seconds later (properly focusing on the distance):

[DateTime                            ] = "2007:06:05 13:12:51"

IMG_1062 was taken four seconds later again:

[DateTime                            ] = "2007:06:05 13:12:55"

The Ty photos

Interestingly (and this wasn’t lost on forum commenters at the time), these photos were handled by Adobe Photoshop CS2 (“Creative Suite” version 2) Macintosh, as is abundantly clear from the metadata. These annotate an image being loaded (at 22:32:53 on 2007-06-16), modified, and then saved out 5.5 minutes later (at 22:38:20). The time zone was -06:00.

      |         <xap:CreatorTool>Adobe Photoshop CS2 Macintosh</xap:CreatorTool>
      |         <xap:CreateDate>2007-06-16T22:32:53-06:00</xap:CreateDate>
      |         <xap:ModifyDate>2007-06-16T22:38:20-06:00</xap:ModifyDate>
      |         <xap:MetadataDate>2007-06-16T22:38:20-06:00</xap:MetadataDate>

The first image I looked at (DroneBigBasinTy060507aa.jpg) had a UUID of 22702A38-1DB6-11DC-8078-C1028E507E7C, which decodes to:

• Date/time = 2007-06-18 16:08:21.330181.6 UTC
• MAC address = c1:02:8e:50:7e:7c (which, once again, is almost certainly randomised per session)

i.e. 2 days after the CS2 date. JPEGsnoop’s overall verdict:

EXIF Software: OK [Adobe Photoshop CS2 Macintosh]
 SW :[Adobe DNG Converter      ]                                 [                ]                  
 SW :[Adobe Photoshop          ]                                 [Save As 05      ]                  

60507bb.jpg had XMP but no UUID:

      |         <xmp:CreateDate>2007-06-16T22:45:28</xmp:CreateDate>
      |         <xmp:CreatorTool>Adobe Photoshop CS2 Macintosh</xmp:CreatorTool>
      |         <xmp:ModifyDate>2007-06-16T22:47:47</xmp:ModifyDate>

60507cc.jpg had both:

      |         <xap:CreateDate>2007-06-16T22:50:49-06:00</xap:CreateDate>
      |         <xap:ModifyDate>2007-06-16T22:52:33-06:00</xap:ModifyDate>
      |         <xap:MetadataDate>2007-06-16T22:52:33-06:00</xap:MetadataDate>

• 5BADA44A-1DEE-11DC-8078-C1028E507E7C
  • Date/time = 2007-06-18 22:50:49.180065.0 UTC
  • MAC Address = c1:02:8e:50:7e:7c 

60507ee.jpg had only XMP:

      |         <xmp:CreateDate>2007-06-16T22:41:33</xmp:CreateDate>
      |         <xmp:CreatorTool>Adobe Photoshop CS2 Macintosh</xmp:CreatorTool>
      |         <xmp:ModifyDate>2007-06-16T22:43:14</xmp:ModifyDate>

60507ff.jpg had only XMP:

          |         <xmp:CreateDate>2007-06-16T22:53:02</xmp:CreateDate> 
          |         <xmp:CreatorTool>Adobe Photoshop CS2 Macintosh</xmp:CreatorTool> 
          |         <xmp:ModifyDate>2007-06-16T22:54:34</xmp:ModifyDate> 

60507gg.jpg had only XMP
          |         <xmp:CreateDate>2007-06-16T22:38:36</xmp:CreateDate>
          |         <xmp:CreatorTool>Adobe Photoshop CS2 Macintosh</xmp:CreatorTool>
          |         <xmp:ModifyDate>2007-06-16T22:40:57</xmp:ModifyDate>

60507hh.jpg had only XMP:

      |         <xmp:CreateDate>2007-06-16T22:43:29</xmp:CreateDate>
      |         <xmp:CreatorTool>Adobe Photoshop CS2 Macintosh</xmp:CreatorTool>
      |         <xmp:ModifyDate>2007-06-16T22:44:44</xmp:ModifyDate>

60507ii.jpg had only XMP:

      |         <xmp:CreateDate>2007-06-16T23:03:21</xmp:CreateDate>
      |         <xmp:CreatorTool>Adobe Photoshop CS2 Macintosh</xmp:CreatorTool>
      |         <xmp:ModifyDate>2007-06-16T23:05:26</xmp:ModifyDate>

60507jj.jpg had both:

      |         <xap:CreateDate>2007-06-16T23:01:06-06:00</xap:CreateDate>
      |         <xap:ModifyDate>2007-06-16T23:03-06:00</xap:ModifyDate>
      |         <xap:MetadataDate>2007-06-16T23:03-06:00</xap:MetadataDate>

• UUID 668A3FCB-1DEF-11DC-8078-C1028E507E7C
  • Date/time = 2007-06-18 22:58:16.899783.5 UTC
  • MAC address = c1:02:8e:50:7e:7c

60507kk.jpg had both:

      |         <xap:CreateDate>2007-06-16T22:58:42-06:00</xap:CreateDate>
      |         <xap:ModifyDate>2007-06-16T23:00:30-06:00</xap:ModifyDate>
      |         <xap:MetadataDate>2007-06-16T23:00:30-06:00</xap:MetadataDate>

• UUID = 668A3FC7-1DEF-11DC-8078-C1028E507E7C
  • Date/time = 2007-06-18 22:58:16.899783.1 UTC
  • MAC address = c1:02:8e:50:7e:7c

Finally, 60507ll.jpg had both:

      |         <xap:CreateDate>2007-06-16T22:55:22-06:00</xap:CreateDate>
      |         <xap:ModifyDate>2007-06-16T22:58:17-06:00</xap:ModifyDate>
      |         <xap:MetadataDate>2007-06-16T22:58:17-06:00</xap:MetadataDate>

• UUID = 5BADA452-1DEE-11DC-8078-C1028E507E7C
  • Date/time = 2007-06-18 22:50:49.180065.8 UTC
  • MAC Address = c1:02:8e:50:7e:7c 

From this, it seems as though these images were initially processed on 2007-06-16 between 22:32 and 23:03, before being saved out two days later (as a batch?) between around 22:50 and 22:58.

The Drone Research Team Forum

According to MUFON, Raj’s 12 drone pictures were sent to Linda Moulton Howe who scanned them in and posted them. So any digital forensic analysis of these should only lead back to her, not to him.

In terms of content analysis, I should note that one group of 2007-drone researchers (the now-defunct Drone Research Team, though their pages live on in the Wayback Machine) believed that they had identified the precise telegraph pole in Capitola, CA that appeared in Raj’s pictures. They even printed out 400 flyers and posted them to all the pole’s neighbours to see if anybody had seen anything. (I believe the answer was a resounding no.)

Incidentally, the Drone Research Team’s main members were (according to this site):

• Tomi01uk (UK)
• Onthefence (Canada)
• 10538 (USA)
• Nemo492 (France)
• Raska (France)
• Elevenaugust (France)

To identify the site, they hired private investigators Frankie Dixon and T.K. Davis, but also asked them to try to identify the other drone sites. (A story covering their search ended up in the Los Angeles Times in 2008.) Here’s an animation created by arkhangels overlaying one 2007 drone image with an image taken by the private investigators:

Even though “Chad” claimed to have taken his photos in Bakersfield, it turned out that the actual location was a little distance away. Similarly, the “Stephen” drone picture turned out to be not Big Basin State Park, but (thanks to Pacific Gas and Electric meter reader Tom) Bohlman Road Ridge in Saratoga.

Who was the mysterious “Isaac“, who claimed to have been working on an alien language in a Palo Alto research institute (“CARET”) in 1984-1987? In 2007, this Isaac posted a page on the free hosting website Fortune City (which has now been archived) with a load of scanned ‘alien’ documents; then answered various follow-up questions (I found what seems to be a complete archive of these on the “Metallicman” website); and then completely disappeared. Everything online since then relating to Isaac’s actual identity appears to be 50% speculation, 50% noise, and 0% fact.

Might a white hat hacker be able to find more details about Isaac, e.g. his IP address, email address etc? I think probably not, because I believe that Fortune City’s account details or server logs were never leaked or exploited (though please tell me if I’m wrong). After a heavily-oversubscribed IPO at the peak of dotcom mania, Fortune City crashed in 2012, and became Dotster (had you ever heard of Dotster? No, me neither). Now, not unlike Ozymandias, “nothing beside remains” of this “king of kings”, and only “the lone and level sands stretch far away”.

All the same, my question today is this: might digital forensics be able to identify “Isaac”?

Under A Digital Microscope

I started by examining Isaac’s JPEGs: these had no metadata or even comments, and their raw data (using HxD) revealed nothing of interest. JPEGSnoop, however, revealed that all the images appeared (according to its database of JPEG header signatures) to have been saved out from Adobe Photoshop. The range of different quality options used suggest to me that the user was (at least) a fairly experienced Photoshop user.

The JPEGs divided into two obvious groups:

Document scans:

• 2550 x 3274 quality 82 – p119- Adobe Photoshop – Save as 07
• 2550 x 3199 quality 90 – p120 – Adobe Photoshop – Save as 09
• 2550 x 3234 quality 90 – p121 – Adobe Photoshop – Save as 09
• 2550 x 3203 quality 90 – p122 – Adobe Photoshop – Save as 09
• 2550 x 3247 quality 90 – p123 – Adobe Photoshop – Save For Web 015
• 2550 x 3298 quality 53 – cover – Adobe Photoshop – Save as 08
• 2550 x 3313 quality 87 – p2 – Adobe Photoshop – Save as 08
• 2550 x 3266 quality 87 – p3 – Adobe Photoshop – Save as 03
• 2550 x 3290 quality 76 – p4 – Adobe Photoshop – Save as 05
• 2550 x 3294 quality 82 – p5 – Adobe Photoshop – Save as 06
• 2550 x 3255 quality 86 – p6 – Adobe Photoshop – Save as 05
• 2550 x 3274 quality 82 – p7 – Adobe Photoshop – Save as 04
• 2550 x 3278 quality 76 – p8 – Adobe Photoshop – Save as 03
• 2550 x 3255 quality 82 – p9 – Adobe Photoshop – Save as 05

The dimensions suggest to me that the scanner’s native resolution was 2550×3300 (or an integer multiple of that, e.g. 1200dpi rather than 300dpi). So I would expect that Isaac used something like an HP Scanjet 3570c, which was a popular choice of scanner at the time (and has a 1200dpi native resolution).

Photographs:

• 1768 x 1203 quality 95 – photo 1 – Adobe Photoshop – Save as 11
• 1768 x 1147 quality 95 – photo 2 – Adobe Photoshop – Save as 11
• 1768 x 1147 quality 95 – photo 3 – Adobe Photoshop – Save as 11
• 1762 x 1151 quality 95 – photo 4 – Adobe Photoshop – Save as 11

Superficially, you might think that the dimensions of these images suggest that they were taken with a digital camera whose native sensor width was 1768 (roughly 2.2MP). However, a web search yielded no obvious technical matches.

Hence it’s far more likely that these were in fact scanned in from 35mm negatives and digitally inverted. What we call “35mm film” is actually made up of a 36mm x 24mm rectangle per individual frame (with a 2mm gap between frames). Hence 36mm = 1.41732 inches, and 1.41732 inches x 1200dpi would yield 1700 pixels, which is tolerably close to 1768 pixels. Note further that 37mm would yield 1748 pixels, so we seem to be very much in the right neighbourhood here.

Finally: I should perhaps also mention that Amped Authenticate offers a set of commercial JPEG analysis tools that seems to be even more turbo-charged than JPEGSnoop, but you (alas) pay handsomely for that privilege.

A Scanner Darkly

What can we tell from the images themselves?

For fun, the first thing I tried was to contrast enhance the areas of the scans that had been redacted, just in case the redaction had been inexpertly done (and the text beneath was still recoverable).

As expected, this produced nothing of interest: but while doing this, I did notice something a little unusual. Even though the source material being scanned was monochrome, a faint streaky blue vertical line artifact appeared about 30% of the way in from the left edge in the scans.

After a little thought, I then realised that this artifact was most likely caused by a flaw in the scanner head itself (which might possibly have been damaged during its manufacture). And I also realised that this could essentially be used as a digital fingerprint for Isaac’s scanner.

Here’s what a raw image looks like in Gimp (at 18.2% of original size):

In Gimp (though you could also use ImageMagick etc), to make Isaac’s scanner’s blue-flaw column visible (it’s between x = 752 and x = 760 in the original 2550-wide images) use the menu option Colors –> Value Invert :

Because of the way JPEG down-samples blocks of colours, the blue column isn’t easily visible in normal images: but once the values have been (numerically) inverted, it becomes clear to see. The redacted text blocks make it particularly easy to see (i.e. it’s visible on black text, but not on white background).

JPEGSnoop helpfully offers the ability to look at individual JPEG planes (the other forensic toolboxes I tried didn’t), so here’s a JPEGSnoop screengrab of the Cb plane for part of the same image, with a patchy vertical white streak where the scanner’s blue artifact is:

This is where the digital forensics chase starts to become interesting…

Once again, “the game is afoot!”

The idea now is simple: even though there may be no direct trace of “Isaac” anywhere on the Internet, might we be able to find any other scans made with his same subtly damaged scanner head and posted online? That is, might we be able to find other scans made with Isaac’s scanner?

Given that Isaac posted his alien writing scans on Fortune City, it seems a reasonable guess that he may well have posted other scans to other free Fortune City accounts.

Furthermore, Isaac’s way of working seems to have been be to leave the width of each scan intact (at 2550 pixels) and to trim its length. So I would initially only be interested in images where the width is exactly 2550 pixels.

Finally, the whole point of Fortune City was that it was a place to host stuff that was completely free (it made its money from banner ads). So we would probably only be interested in 2550-pixel-wide images with this specific blue colour flaw that were also hosted by Fortune City.

Step 1 could be to webcrawl the fortunecity.ws archive (there must surely be a list of accounts?) and compile a list of 2550-pixel-wide JPEGs/JPGs. Step 2 would be to grab them (into an AWS bucket?) and run an image filter on them. Step 3 would be visual inspection, or an automated sort based on a metric.

So… who wants to help give this a go? Will this reveal Isaac’s identity?

Since posting up some images way back in 2009, I haven’t really covered the (allegedly) ‘alien’ language claimed to have been stolen from a (fictitious) “CARET” research institute in Palo Alto by a mysterious Fortune City poster called “Isaac”. The whole lot was – in my opinion – nothing more than a Ufologist-trolling hoax (albeit one of the better-looking ones).

Starfire Tor

However, I recently found out that Isaac’s alien alphabet had (supposedly) been debunked by an online poster called Starfire Tor. She had noticed that the same font had been used by Alienware for a viral-style marketing campaign, based around a competition where breaking a ciphertext could have won you a trip to New York City worth $2800. Here’s what the ciphertext looked like (image from Starfire Tor’s website):

Alienware (which by then had been Borg-ed up by Dell Computers) also used the font to stamp “ALIENWARE” onto their promotional desktops (image also from Starfire Tor’s website):

For Starfire Tor, this was a slamdunk: a huge corporation like Dell would never (she reasons) just steal someone else’s font, ergo Dell/Alienware must have commissioned the font design in the first place, ergo they must have been (somehow) behind the whole Isaac/CARET thing. End of story.

However… take a closer look at all three versions of the alien alphabet, and you’ll notice they’re all slightly different. The competition alphabet contains four extra glyphs (plus a dash and a full stop) not in the Isaac alphabet: while the ALIENWARE stamped-out alphabet has one of the new competition glyphs (for the “A” in “ALIENWARE”) plus a unique reflected version of a glyph in the Isaac alphabet (the “E” in “ALIENWARE”). Additionally, the three alphabets all render the alien vertical bar glyph in different ways.

Hence it seems as though what actually happened was that Dell/Alienware just got their in-house artists to rip off the bloody aliens. (Presumably hoping that they came in peace, rather than with trademark attorneys?) So, even though I’m sure Alienware founder Nelson Gonzalez (who was famously a fan of all things ufological, hence his company’s name) would have loved to have been part of an alien conspiracy, I don’t believe that this was what happened (or else Dell would have just reused the existing font, right?)

Anatomy of an Alien (Alphabet)

In my 2009 post, I noted that it looked as though the alien text was made up of some letters, some numbers, and a few pieces of what seemed to be punctuation. I also complained that nobody had actually bothered to transcribe the alien text (presumably because going round in circles is a pain in the neck).

All the available Isaac CARET scans are online here, taken from pages 119 to 123 of a fictitious CARET book. Note that pages 120 to 123 are just zoomed-in versions of parts of the (larger) diagram on page 119, so there’s actually only a single diagram to work with.

Looking more closely, the alphabet contains a large number of apparent groupings, which suggest that a kind of “pigpen”-style glyph generation process might have been in play here. With that in mind, here’s my work-in-progress transcription key for Isaac’s alien alphabet:

It’s entirely possible that I’ve missed one or two really rare glyphs (the lettering is large in some places and tiny in others), but I believe that this covers just about everything that appears. (I’m reading all the strings clockwise.)

Alien Alphabet Transcription

Inevitably, I tried to use this to transcribe some of the text (in the middle of the “octal junction”, p.120):

FKRYRSAKML SBUN M HY

X2L R -JM EW1D DT-ED (345-521) BV-KA P6FKL (])

SHJD C-XEGYRI (DEB)

JMRI LAI-FELK GUHFVX (KLN) [

However, I have to point out that CryptoCrack wasn’t hugely impressed. But maybe someone else will have more perseverance and luck than me.