Summary

  • Pietrosemoli's own 2008 paper prints 382 km in the body and "381 km path" in a figure caption, prints 279, 280 and 282 km for the El Aguila–El Baúl path, and leaves the 6 Mbps figure attached to TIER TDMA routers on a link it calls unstable.
  • The geography of the record can be checked from outside; the rate cannot, because no document read prints the antenna gain at both ends of the 382 km test, names the TIER router model, or states an averaging interval.

Ermanno Pietrosemoli's account of the Venezuelan long-distance links survives in three usable shapes: the 2008 journal paper (JoCI article record, author-hosted text), a coauthored ACM workshop paper the same year (NSDR '08), and a workshop deck from February 2009 (Long Distance Links). A full-text mirror of the journal paper (Distant Reader copy) reproduces the body passages. What follows is read from retrieved extracted text, so exact wording is quoted as it appears in that text; the publisher's article body was not separately page-opened.

Two places, and a number that will not sit still

The 279 km experiment ran between Pico del Aguila and El Baúl, the eastern end at cerro Morrocoy. In the same paper the path is also called 280 km, and a link-budget calculation uses 282 km. None of those three is presented as a correction of the others. The deck's timeline fixes the year as 2006 and shows the site using a 2.4 m dish and a 2.7 m mesh reflector.

The headline test is the other pair: El Aguila to Platillón. The abstract says experiments "at distances of up to 382 kilometer" in Venezuela, and prints the window as "April 206 to July 2007" — the year as typed, not silently repaired here. The deck dates the 382 km work to April and August 2007. The body of the paper says 382 km; the figure caption beneath that same section reads "381 km path".

My own subtraction, and nothing more than that: the two figures describe one path and differ by a kilometre. It does not tell you which is right, and it does not turn a caption into a measurement.

The equipment column, read line by line

The paper's most useful habit is that, for the first link, it prints the bill of materials. The radios were a pair of Linksys WRT54G routers, one as access point and one as client, running OpenWRT so the acknowledgement timing could be stretched for distance. Output was set to 100 mW, 20 dBm; the text says pushing to 200 mW was undesirable because of nonlinearity and spurious signals. The antennas were recycled parabolic dishes, the 2.75 m mesh estimated at about 32 dBi by comparison against a commercial 24 dBi unit — an estimate the paper itself hedges, noting uncertainty from reflected signals.

Then the paper states a design requirement: 30 dBi antennas at both ends for the roughly 282 km path. That is a requirement, not a printout of what every antenna actually was.

The upgrade is described with equal care inside its own section. The TDMA repeat used commercial 30 dBi antennas and routers modified by the TIER group; measured throughput was 3 Mbps in each direction, 6 Mbps total, matching simulations the paper attributes to the TIER team at Berkeley. Before that upgrade, the Linksys link moved PDF files at about 65 kbps.

And then the equipment column thins out. For the 382 km test the paper says "Again" the Linksys and TIER-supplied routers were used. Across the five documents read here, it restates neither the antenna gain for that link nor the TIER router model. The 30 dBi sentence belongs to the 279 km repeat. Anyone who writes "30 dBi at 382 km" is borrowing a figure from one paragraph and a distance from another.

What the 6 Mbps is actually sitting on

The paper never claims the record was a service. At 382 km the Linksys equipment showed about 1% packet loss and 12 ms average round-trip time, the TIER equipment no packet loss — and the link was not stable: signal fluctuations often interrupted communication, and the 6 Mbps total bidirectional was measured only when the received signal was sufficient. The conclusion is blunter than the abstract. It calls the 280 km link stable, and calls the roughly 380 km link "probably at the edge of the 2.4 GHz link capabilities with low cost equipment".

The deck compresses all of that into a timeline entry: "382 km, 6 Mbit/s". It does add one thing the paper does not say so plainly: both acknowledgement-timing modification and TDMA were tried at 382 km, and TDMA produced a two-orders-of-magnitude improvement in throughput.

What no reviewed document supplies is the interval. No averaging window, no test duration and no date on which the 382 km stability was last assessed appear beside the 6 Mbps figure. A reader can see the qualifying sentence and still cannot compute a sustained rate from it.

Where the arithmetic stops

Here is what a stranger can do with these documents and no equipment. Name the endpoints. Compare the printed distances for one path and see the discrepancy. Reconstruct which radio and which antenna each paragraph attaches to, and see that the strongest specification sits with the shorter path. Read the author's own stability caveat beside the deck's compressed version of it, and note that the compression drops the caveat.

And here is what a stranger cannot do. Confirm the gain at either end of the 382 km link. Confirm the TIER router model. Confirm what "6 Mbps" was averaged over. My own restatement, labelled as mine: 6 Mbps against 65 kbps is a factor of about 92, and that number describes an upgrade across different hardware, not a service level anyone experienced.

That is the finding, and it is not a scandal — it is a shape. The number easiest to repeat is the distance, and the distance is the part that is genuinely checkable. The number that travels fastest in slides is the rate, and the rate is the part the author himself confined to sufficient-signal conditions. Pietrosemoli printed the qualification next to the claim. Two decades of retelling kept the number and dropped the sentence.