PDA

View Full Version : SETI@home yields to pressure to curb cheating


gypseyman
5th November 2002, 03:58 PM
By Andrew Colley, ZDNet Australia
04 November 2002

Administrators of the alien-hunting distributed computing experiment SETI@home have announced they will crack down on cheats who rort statistics on computing power lent to the project. The announcement follows a united protest from the chief contributors.
SETI@home director David Anderson announced SETI@home would do its best to investigate users returning suspiciously high amounts of work and delete their accounts if it uncovered solid evidence of cheating.

The announcement comes after over 800 of the projectís keenest contributors signed a petition demanding that SETI@home administrators act to curb the rorts.

The petition was organised last week after veteran "cruncher" and cheat-hunter, Max Nealon, exposed a massive increase in the amount of illegitimate work units being returned to SETI@home.

Nealon said he approached the media because he believed that cheating was dominating the work returned by the projectís top contributory sources, leading to a vast overstatement of the projectís power.

"When talking to sponsors about their four million users Ė 200,000 active, 4,000 producing more than 90 percent of the work -- they are over-stating their power," said Nealon.

Nealon revealed startling statistics of cheating from SETI@Netherlands team leader Rick Groenewegen who estimated that 41 percent of around eight million work units returned by the team were produced by a handful of miscreants that had infiltrated the team.

Groenewegen, who has sinced re-confirmed the statistic' accuracy said, the situation was very disappointing as the vast majority of the team were honest crunchers.

According to Groenwegen there have been many instances of cheating throughout the projectís three-year history, and across many teams.

Groenwegen said that the community behind the project was starting to grow weary of being ignored by Berkeley and that the latest revelations had forced people to rethink their decisions to continue contributing to the project.

"People were getting sick and tired of Berkeley ignoring requests from the community," said Groenwegen. "There were indeed people that were willing to go to other projects where the staff would listen to them".

Some users have trivialised the cheating problem as a storm in a tea cup. They say that the project's organiser should remain focused on preserving the integrity of the experiment's scientific method not chasing cheats, however Groenwegen bemoans that reasoning.

"It is true that the cheats that were used by the 'recent' cheat[s] didn't affect the science, but projects like SETI@home are powered by the effort of the volunteers.

"If those volunteers think it's a big problem than Berkeley should consider it a big problem too," he said.

When asked why SETI@home had decided to act on the problem now Anderson said that he just had found some "spare time" to work on it.

"We know that individual and team statistics are very important to many of our users, and we'd like to keep accurate statistics," said Anderson in his statement.

Anderson said last month that SETI@homeís project designers were in the midst of finalinsing a deal to secure new funding.

But Anderson said that the administrators were still hampered by man-power problems and said addressing the cheating problem would have to be prioritised against the rest of the projectís goals.

For now he has asked contributors to be patient and asked for their assistance to make SETI@homeís successor, Berkeley Open Infrastructure for Networked Computing (BOINC), more cheat-resistant.

"Please bear with us until BOINC is released. BOINC is open-source, so programmers can help us by reviewing its cheat-prevention system," said Anderson.

Some users have called on SETI@home to give team leaders more power to manage accounts. However Groenwegen believes this would be a band-aid measure that would place too much responsibility on the team leaders. He believes structural changes need to be built in to the server software.

"It would be helpful, but my opinion is that if you do that you're not working for a solution. If you add an option like that, it's just a 'quick-fix'," he said.

jema
5th November 2002, 04:57 PM
about time :) the cheats undermine the fun of the project for the vast majority and that undermines the project itself.

jema

Chevy_Avalanche
5th November 2002, 05:05 PM
My thoughts exactly.

doing Seti@Home Work Units for Berkeley and belonging to a team and chatting with the is always fun and it should be always be that way

Gservo
5th November 2002, 05:26 PM
Yeah, it is fun, and should not be spoiled by those just seeking ego trips