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 In 2007, this Isaac posted a page on the free hosting website Fortune City (which has now been archived) with a load of scanned ‘alien’ documents; then answered various follow-up questions (I found what seems to be a complete archive of these on the “Metallicman”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?
Mr. Pelling:
In the line ‘2550 x 3274 quality 82 – p119- Adobe Photoshop – Save as 07’,
can you please tell us what the ‘Save as 07’ indicates?
leifFraNorden: for exporting images as JPEGs, Adobe Photoshop’s user interface offers a range of quality settings (e.g. quality “01”, quality “02”, quality “03” etc), which it then converts to JPEG quantization tables and other settings in an exported JPEG file’s header. JPEGSnoop contains its own database of more than 3000 such JPEG settings, which it compares against the a given file’s JPEG headers to determine which application it thinks wrote out the JPEG.
You could look at reviews on amazon within the required timeframe – say an engineer with an interest in photography.
https://www.amazon.com/HEWLETT-PACKARD-HEWQ2707A-ScanJet-Scanner/product-reviews/B000067R8X
By way of example – not saying this is the guy…
An engineer named E. Siegal – reviewed products obsessively on Amazon (tick in the box)
Person with the same name claims to be a genius inventor (tick in the box).
He has handwriting samples on his web page (you could compare handwriting samples with e..g. freely available German HAT software) with the alien texts.
So someone ‘like this’ who thinks his genius was overlooked/exploited and posts about it and now decides to fool the world?
But makes the mistake to show his handwriting…
Since the date in isaaccaret1.fortunecity.ws is ‘June 2007’, I’m wondering why an Andrew Johnson wrote in his praising post from 1 July 2007, that they had first appeared ‘a couple of months ago’.
https://www.checktheevidence.com/wordpress/2007/07/01/old-news-2852/
Thomas: the original “sightings” (which preceded the Isaac/CARET Fortunecity page) were of supposed ‘drone’ images with broadly the same alien font. I plan to post about those drone images separately.
This site is entertaining… and points out Isaac’s issue re. parental divorce.
http://www.theironskeptic.com/articles/issac/issac.htm
Here is our speculation and noise concerning Issac and his hoax. We sincerely hope you find some fact somewhere, and that at least some of our noise sounds informed. Evidence first:
1. THE CHAD DRONE IMAGES– [https://www.coasttocoastam.com/pages/chad-drone-images/] ‘Chad’ emailed these to the Coast to Coast website in 2007. Look at the first image, showing the ‘drone’ behind a blossomed tree. The contrast between the lights and darks on the leaves is high, yet the drone appears in low contrast. The sun is the only light source and is consistent throughout, but tree and drone are lit differently.
In our experience, early digital cameras did not handle contrast in bright sunlight well at all– an issue known as dynamic range. Somehow the drone seems entirely immune to the limits of digital camera sensors.
Look at the third image, showing the drone over the treetops, both should be equally in focus, but the drone appears sharper than the treetops.
Several images (1, 3, and 5) show the drone partly obscured by branches. The drone/branch intersection is minimal– just enough to add a sense of realism. An intersection involving more pixels would be time consuming and difficult to pull off. The images are low-res (the first: 792 x 519), which would hide flaws in the Photoshop.
Image 4 shows a mountainous forest. This together, with the location of similar photos from ‘the Lake Tahoe witness’, make it likely the location shots were taken in California.
Conclusion: ‘Chad’s’ photos are PhotoShopped composites, with the drone probably being a computer generated image. The same characters appear in the fake drone images and in Issac’s CARET ‘Research Report’, indicating Chad, et. al. and Issac are one and the same. Falsus in uno, falsus in omnibus.
2. THE ‘ANTI GRAVITY DEVICE’– [https://isaaccaret.fortunecity.ws] This is a digitized photograph of a 3d model. Look at PACL Q4-86 Report Photo 4.1. The focus is on the front of the object, with the background slightly out of focus. (Look at the front and back wheels at 200%.) This is typical of a model, not a 3d computer image. A tooth on the top of the second ‘dial’, appears slightly bent– which virtually proves the object is a model.
There is evidence of photographic grain in the shadow under the object, which you see when scanner resolution is higher than the grain of the film. Digital photography replaced film rapidly in the early 2000s– by 2004 film had practically disappeared in commercial photography. Therefore, we conclude the object was photographed before 2004.
The lighting is quite professional– it looks like the model was shot by an experienced photographer using professional equipment– though the background has been retouched in Photoshop.
The model, like the photography, is of professional quality. 2004 (or even 2007) is a bit early for 3d printing, so we suspect the object uses repurposed parts (but from where?). It looks like the ‘alien’ characters are applied as decals.
Conclusion: The anti gravity device is a model, photographed before 2004.
3. THE ‘LINGUISTIC ANALYSIS PRIMER’– [https://isaaccaret.fortunecity.ws] This diagram is created using an object-list (‘vector’) application. Commercial software in this class includes Adobe Illustrator, CorelDRAW, and (now defunct) Macromedia Freehand. The diagram is highly complex– meaning the file size is very large– and the lines connecting the elements incorporate Bézier curves. The ‘alien’ font looks composed from more simple curves.
Regarding the diagram: Issac claims: [https://metallicman.com/laoban4site/the-disclosure-of-the-caret-program-at-pacl] ‘We copied it into a drafting program over the course of about a month. Our software was understandably primitive by today’s standards, but it was still orders of magnitude more powerful than a pencil and paper would have been.’
Issac claims to have been at PACL between 1984 to 1987. At that point, he would have needed a fairly powerful computer to have produced so complex a diagram. But the first graphic editor to incorporate Bézier curves was Adobe Illustrator, released in 1985. [https://www.vectornator.io/blog/bezier-curves/] The problem was that Illustrator ran on an Apple Macintosh computer, which did not have the memory or processing necessary to produce such a complicated diagram.
Tellingly Issac avoids telling us which software he used, but we know of no electronic publishing or CAD system capable of producing such a diagram in the mid 1980s. (As we used a number of the earlier systems, we should know.)
Conclusions: The linguistic analysis primer was produced no earlier than the early 1990s, and Issac is not as familiar with early graphics systems as he pretends to be.
4. ANALYSIS– An obvious hoax, but an interesting one. The ‘alien alphabet’, antigravity device, and linguistic analysis primer were created between 1993 and 2003, several years before the hoax was perpetrated. The quality of design is professional (the primer in particular is quite beautiful) and took a considerable amount of time to complete.
The antigravity device doesn’t match the description in the CARET report: ‘Their hardware appeared to be perfectly solid and consistent in terms of material from one side to the other. Like a rock or a hunk of metal.’ The relationship between the primer and the report is less than plausible.
Most hoaxes are much simpler– consider Billy Meier’s UFO photos, the Patterson-Gimlin bigfoot film, or the McMinnville UFO photos. Why was there so much time put into this one? why did several years pass before the material was published? Why don’t the designs align better with the report?
We guess that the designs are repurposed, and began with some different goal– perhaps as a personal project. The antigravity device photos look like product shots– as if Issac hoped to market the object, perhaps as a toy or a desk novelty. The alien alphabet as well could have fit with a sci-fi product line. When this did not come to fruition, Issac only later repackaged them as a hoax.
Whoever photographed the model really understands light in photography, but there is no evidence for this in the Chad Drone Images– so it looks like a second party produced the photos.
It is quite possible that the alien alphabet does not encode anything meaningful. Because the linguistic analysis primer is essentially abstract art, it requires a leap of faith that the alien alphabet strings carry a code. If they do the meaning is probably not enlightening and possibly playful.
Issac worked as a (probably graphic) designer in California sometime between 1993 and 2003. He has an interest in Sci-Fi and the history of computer graphics and is an independent inventor type. He has probably shown his work to co-workers and friends.
5. FINAL WORD– Issac has not profited from this enterprise, and ‘Coast to Coast’ is not where you go if you want to be taken seriously. Perhaps Issac intended his publication as a postmodern literary endeavor rather than a hoax.
We kind of hope the game is not afoot. Issac seems a bit like Charles Dellshau, and CARET a bit like the Sonora Aero Club. We appreciate the artistic endeavors of both. It strikes us that Issac may well be neurodivergent (a term we hate)– if this is true, identifying Issac may hurt him.
James: it’s a light, snarky take on the whole Isaac thing – but doesn’t really manage to “jump the snark” to anything substantial on the other side. The whole “bright 14-year-olds with parent issues trying to gain attention by pretending to be someone older” thing is pretty much Reddit in a box, so it’s not really casting as strong a raking light on the mystery as it might.
LeifFraNorden: lots of good insight and ideas there, thanks! I plan to post separately about the drones (specifically the use of the alien alphabet on the drones), though I would note that I’m not 100% convinced that items necessarily date to before 2007.
nickpelling: Yes, the model could have been shot on film after 2004. But in that case Issac made quite a bit of extra effort to produce an effect that was not central to his hoax. Look at image PACL Q3-85 Inventory Review p56– which has been laserprinted and photocopied (perhaps multiple times). There is no way of telling whether the original was digital or film.
Also note that the model was photographed with considerable care, but the drone photos were created with much lower standards. It doesn’t appear they were created at the same time.
We have done this kind of work ourselves. The long hours wear one down. From our experience we believe with moderate confidence the slides were shot when film was still the standard.
LeifFraNorden: I understand and appreciate your conclusion, in the end these things are more a weighting of evidence than an absolute position. I have some additional stuff to add in the next post which I’m sure you will find interesting…
I don’t know if it might help, but you can find the original crawled links with the wayback machine.
https://web.archive.org/web/20080218153605/http://isaaccaret.fortunecity.com/